Psychology A Study Of A Science Volume 3 Formulations Of The Person And The Social Context

150

v.3 63-15383
Psychology: a study of
a science*

K AN S AS C IT Y MO . PU BLI C LI : B R A R V

MM MAY 1 A 1991

Psycholgy: A Study of a Science

STUDY I. CONCEPTUAL AND SYSTEMATIC

Volume 3. Formulations of the Person
and the Social Context

PSYCHOLOGY: A STUDY OF A SCIENCE

The Series

STUDY I. CONCEPTUAL AND SYSTEMATIC

Volume 1 . Sensory, Perceptual, and Physiological Formulations

CONTRIBUTORS: Albert A. Blank, James J. Gibson, C. H. Graham, D. 0. Hebb,
Harry Hehon, J. C. R. Licklider, Clifford T. Morgan, Kenneth N. Ogle,
M. H. Pirenne and F. H. C. Marriott, Leo Postman and Edward C. Tolman,
W. C. H. Prentice

Volume 2. General Systematic Formulations, Learning, and Special

Processes

CONTRIBUTORS: Dorwin Cartwright, Douglas G. Ellson, W. K. Estes, F. C.
Prick, Edwin R. Guthrie, Harry F. Harlow, R. A. Hinde, Arthur L. Irion,
Frank A. Logan, Neal E. Miller, B. F. Skinner, Edward C. Tolman

Volume 3. Formulations of the Person and the Social Context

CONTRIBUTORS: Solomon E. Asch, Raymond B. Cattell, Franz J. Kallmann,
Daniel Katz and Ezra Stotland, Paul F. Lazarsfeld, Henry A. Murray,
Theodore M. Newcomb, Talcott Parsons, David Rapaport, Carl R. Rogers,
Herbert A. Thelen

STUDY II. EMPIRICAL SUBSTRUCTURE
AND RELATIONS WITH OTHER SCIENCES

(These titles in preparation)

Volume 4. Biologically Oriented Fields: Their Place in Psychology and in
Biological Science

Volume 5. The Process Areas, the Person, and Some Applied Fields: Their
Place in Psychology and in Science

Volume 6. Investigations of Man as Socius: Their Place in Psychology and
the Social Sciences

POSTSCRIPT TO THE STUDY

Volume 7. Psychology and the Human Agent: A View of Problems in the
Enaction of a Science (by Sigmund Koch]

Psychology: A Study of a Science

STUDY I. CONCEPTUAL AND SYSTEMATIC

Volume 3. Formulations of the Person
and the Social Context

Edited by Sigmund Koch

DUKE UNIVERSITY

McGRAW-HILL BOOK COMPANY, INC.

New York Toronto London

1959

PSYCHOLOGY: A STUDY OF A SCIENCE was made possible
by funds granted by the National Science Foundation to the Amer-
ican Psychological Association, and carried out under the spon-
sorship of the latter organization. Neither agency, however, is to
be construed as endorsing any of the published findings or con-
clusions of the Study,

Copyright © 1959 by the McGraw-Hill Book Company, Inc. Printed
in the United States of America. All rights reserved. This book,
or parts thereof, may not be reproduced in any form without permis-
sion of the publishers. Library of Congress Catalog Card Number 57-14691

in
35273

THE MAPLE PRESS COMPANY, YORK, PA.

PREFACE

When one looks back over the history of science, the successes are
likely to be stressed and the failures forgotten. Thus one tends to see
science as starting with a sure sense of direction and progressing neatly
to its present form. Or so it is for the older and well established
branches of science; but not for psychology. Psychology has not one
sure sense of direction but several quite unsure directions. Growth is
erratic and there is much casting about for the most crucial problems
and the most powerful methods. These apparent differences between
psychology and the older branches of science may result from the
djfficul^^ it is perhaps significant that

many of the problems of psychology were not attacked by the methods
of science until so late a date in history. Or the differences may be an
illusion resulting from the much closer view we have of the beginning
struggles to develop a science of psychology than we now have of the
beginning efforts in the older sciences.

Certainly psychology has its problems, and they are not easy.
Nevertheless, knowledge has grown rapidly in the short history of
man's efforts to develop a science of behavior, and the time seems ap-
propriate for a major effort to examine the progress that has been
made in attempting to find a way, or ways, to the attainment of the
explanatory power that we like to think of as characteristic of science.
A growing body of empirical information, a serious concern over
methodological issues, and a variety of efforts to bring a selected body
of fact into the organizing framework of theory all emphasize the need
for that line of questioning — always going on in science — which
explores the shape of knowledge, the range and inner connections of
the ideas through which it has been developed and organized, the
changing substructures of empirical data, and their emerging relations
to each other and to the findings of other sciences. The seven volumes
of Psychology: A Study of a Science are a response to this need.

The first three volumes, which bear the collective title Study L
Conceptual and Systematic, are concerned with many of the systematic
formulations of recent and current influence which psychologists have
developed to account for the phenomena in which they are interested.

Vi BAEL WOLFLE

Each systematic position is analyzed by its originator, or a person con
nected with its development, in a way which gives attention to th
problems it seeks to solve, the empirical basis on which it rests, it
degree of success, and its relations to other formulations.

A second set of three volumes, collectively called Study II. Empirica
Substructure and Relations with Other Sciences, inquires, again througl
the efforts of creatively active investigators, into the organization o
various fields of empirical knowledge, the relations of one to another
and to work going forward in other sciences. It also examines suet
problems in reverse through the participation of social and biologicaJ
scientists who consider the relations of their own special fields to vari-
ous parts of psychology. The three volumes of Study II, now in prepa-
ration, will be published at a later date.

Volume 7 — Psychology and the Human Agent — will present the Study
Director's view of certain problems of psychological inquiry in the light
of the findings of the project.

Primary credit for the initiation of these studies goes to the Asso-
ciation's Policy and Planning Board, which decided in 1952 that the
time had come for a thorough and critical examination of the status
and development of psychology. The National Science Foundation
agreed upon the desirability of such an undertaking and has gener-
ously supported the effort. When funds from the National Science
Foundation were found to be insufficient for all of the expenses of the
studies, the American Psychological Association provided the sup-
plementary funds necessary to complete the work.

From the beginning, the study was divided into two parts. One part
dealt with the education of psychologists and the factors conducive to
research productivity in psychology. That part was directed by Profes-
sor Kenneth Clark of the University of Minnesota, who has reported
the findings in America's Psychologists: A Survey of a Growing Profession,
published by the American Psychological Association in 1957.

The other part, the part with which the present series of volumes is
concerned, has dealt with the substance of psychological thought and
data. Professor Siginund Koch of Duke University has been responsible
for this part of the study. Working closely with him has been a panel of
consultants consisting of Lyle H. Lanier, Howard H. Kendler, Conrad
G. Mueller, and Karl E. Zener. These men, but chiefly Dr. Koch, have
planned, organized, interpreted and edited the work, and successfully
enlisted the cooperation of the approximately 80 authors whose origi-
nal papers will constitute the basic material of the series.

In the background, at a safe distance from the labors that have
sometimes engulfed Dr. Koch, his panel of consultants, and the pri-
mary authors, has been a steering committee on which I had the pleas-

Preface vii

ure of serving as chairman, and having as colleagues Clarence H.
Graham, Lyle H. Lanier, Robert B. MacLeod, Eliot H. Rodnick, M.
Brewster Smith, and Robert L. Thorndike. The steering committee
helped to make administrative arrangements and helped to decide
on the scope of the studies, but takes no credit for their successful
completion.

In the preface to Americans Psychologists we have already acknowl-
edged our gratitude to Kenneth Clark and his collaborators who
helped to produce that volume. It is our final pleasant duty to express
our thanks to Duke University for making Dr. Koch's time available;
to the National Science Foundation for its necessary and generous
financial support and for the counsel and support of John T. Wilson,
Assistant Director for the Biological Sciences; to Lyle H. Lanier,
Howard H. Kendler, Conrad G. Mueller, and Karl E. Zener for their
critical and devoted help; to all of the authors whose names appear on
the title pages for their original contributions; and — most of all — to
Sigmund Koch for directing and driving through to completion what
we hope will be an oft-consulted aid to the scholars and research
workers who are striving to increase the rigor and further the develop-
ment of scientific psychology.

Dael Wolfle, CHAIRMAN

STEERING COMMITTEE

POLICY AND PLANNING BOARD

CONTENTS

Preface v

Dael Wolfle

Introduction to Volume 3 1

Sigmund Koch

Preparations for the Scaffold of a Comprehensive System . . 7

Henry A. Murray

The Structure of Psychoanalytic Theory: A Systematizing At-
tempt 55

David Rapaport

A Theory of Therapy, Personality, and Interpersonal Relation-
ships, as Developed in the Client-centered Framework , 184

Carl R. Rogers

Personality Theory Growing from Multivariate Quantitative Re-
search 257

Raymond B. Cattell

Psychogenetic Studies of Twins 328

Franz J. Kallmann

A Perspective on Social Psychology 363

Solomon E. Asch

Individual Systems of Orientation 384

Theodore M. Newcomb

A Preliminary Statement to a Theory of Attitude Structure and

Change 423

Daniel Katz and Ezra Stotland

Latent Structure Analysis 476

Paul F. Lazarsfeld

ex

X CONTENTS

Work-emotionality Theory of the Group as Organism . . . 544

Herbert A. Thelen

An Approach to Psychological Theory in Terms of the Theory of

Action 612

Talcott Parsons

Appendix

Suggested Discussion Topics 713

Note on the Use of Discussion Topic Index Numbers . . . 724

Some Trends of Study I (Vols. 1-3)

Epilogue 729

Sigmund Koch
Significance of Sensory Psychology for Certain Methodological

Problems 789

Conrad G. Mueller

Name Index 803

Subject Index 813

INTRODUCTION TO VOLUME 3

Psychology: A Study of a Science is a report of inquiries into the status
and tendency of psychological science. Some eighty distinguished authors
have contributed sustained essays which consider: (Study I) major
theoretical formulations of recent importance; and (Study II) the struc-
ture, mutual interrelations, and associations with other sciences of the
main empirical areas in which psychological research is pursued. The
findings of Study I Conceptual and Systematic comprise the initial
three volumes of the series; Study II Empirical Substructure and Rela-
tions with Other Sciences is reported in the following three volumes. A
final volume by the Study Director — Psychology and the Human Agent
— includes commentary on the significance of the findings.

The present volume is the third in the series and is part of Study I.
Each of the eleven essays in this book is a self-contained presentation
which may be read with profit independently of the others, or of the con-
tents of other volumes. Yet the reader must bear in mind that the present
volume is the receptacle of a fragment of Study I, and that Study I, in
turn, is part of a larger enterprise having certain unifying values, aims,
and methods. For a conception of these latter, the reader is referred to
Dael Wolfle's Preface and to the two detailed introductory sections in
Volume 1 — one for the series, the other specifically for Study I.

As an immediate aid to the reader, a resume of the Study's design is
given.

Study I Conceptual and Systematic. This study involved the in-
tensive analysis of thirty-four "systematic formulations" of widely vary-
ing type and subject-matter reference and all of established influence
in recent psychology. A "systematic formulation" was defined quite
generally as "any set of sentences formulated as a tool for ordering
knowledge with respect to some specified domain of events, or further-
ing the discovery of such knowledge" : in applying this definition, care
was taken that no formulation be precluded by nonconformity to stand-
ardized conceptions of the nature of "theory." Since each systematic
formulation is the end-product of a human effort to see and state order
in a given domain, each analysis was made either by the originator (s)

1

2 SIGMUND KOCH

of the formulation in question or (in a few cases) by individuals crea-
tively associated with the development of formulations of which they
were not the primary authors.

Each systeniatist was invited to approach his work with certain
common themes of analysis in mind. These were designed to invite a
convergence of insight on those problems of systematization which had
emerged from the practice of the past three decades, more or less. Some
of the suggested problems had been conspicuous in previous "meta-
systematic" discussion, but required in our opinion exposure to a wider
range of systematically schooled sensibilities. Others were problems that
seemed critically posed by recent systematic work, yet ones which had
received little or no explicit attention.

The dominating hope was for analyses that might illumine the
relations between the creative processes of systematizing and their publicly
expressed products. It was thus hoped that the atmosphere of the study
might encourage as much concern with background influences, orienting
presuppositions, and working methods, as with conceptual content, re-
search achievements, and prospects. It was felt that analysis of this order
could itself have creative consequences; reflective scrutiny of the extent
and depth envisaged means rethinking. The primary intent of the dis-
cussion themes (and indeed the constant aim of all editorial effort) was
to realize an atmosphere that might invite such emphases. Authors were
requested to make explicit reference to the themes in their writing only
to an extent they deemed appropriate or congenial. The use of the
themes for facilitating the collation of findings was thus a secondary, if
still important, aim. As matters turned out, most authors adhered to them
sufficiently to give the reader an excellent purchase for the detection of
similarities and differences on key issues.

The grounds for the selection of the thirty-four formulations in-
cluded in Study I are given in Volume 1 (pp. 21-27). The aim was
a reasonably balanced diversification of formulations (as judged by
many consultants) with respect to (a] subject-matter reference, and (b)
conceptual and methodological "type," Many significant formulations
that we would have wished to represent in the original list were excluded
by spatial and other arbitrary restrictions. Nor are all formulations
originally chosen included in the present volumes. Though the proportion
of inclusions is remarkable, there were some individuals who could not
participate. We do not, then, claim "representativeness" even in an in-
formal and impressionistic sense. We do, however, claim sufficient di-
versity to extend markedly the range of formulations which in recent
years have been given sustained analytic attention.

Study II Empirical Substructure and Relations with Other
Sciences. This study seeks increased understanding of the internal

Introduction to Volume 3 3

structure of psychological science and its place in the matrix of scientific
activity. Over forty contributors, having distinguished research back-
grounds in psychology, or in related biological and social sciences, were
invited to write papers which examine the organization of empirical
knowledge within subareas of these disciplines, and which chart their
cross connections. Psychologist contributors consider the relations be-
tween their own fields of special competence and the rest of psychology,
and inquire also into relations with relevant segments of other sciences.
Social and biological science contributors examine the relations between
their own fields and psychology.

All authors are individuals whose research interests have bridged
conventionally discriminated fields of knowledge. Each was asked to
place special emphasis on those "bridging problems" which had been
central in his own research experience. As in the case of Study I, an
attempt was made to encourage differentiated and stratified analysis,
and to invite a convergence of visions on significant cross-cutting issues,
by proposing certain common themes of analysis. The "themes" for
Study II comprise a detailed breakdown of the senses in which questions
of "mapping" subject-matter structure and exploring field interrelations
might be entertained.

Though the topography of a science is too vast and labile for com-
prehensive or final mapping, this very fact makes it more important
to assay the contours of knowledge as best we can. Study II exploits
the only resource available for such problems — individual vision — but
in a novel way. It assumes that a pooling of expert, specialized sensibilities
can give insight into the emerging structure of a science of a sort not
ordinarily available.

A fuller statement of the plan for Study II appears in the General
Introduction to the Series (Vol. 1, pp. 1-18). An adequate account
of working methods must await publication of the completed study.

Psychology and the Human Agent. This volume is a postscript to
the Study, representing certain views formed by the Director in its
course. The book (a) records those attitudes towards a science and
science which necessarily color the spirit of the Study, (6) constructs
trends from the massive findings of the two group studies, and (c)
considers, in the light of the Study's premises and apparent trends,
certain problems of psychological inquiry suggested by the practice of
the past several decades.

In this day in which "self studies" and reductions of enigma by
seminar are becoming commonplace in social science, it may be helpful
to mention a few of the special features of Psychology: A Study of a
Science. May we stress (in random order) the following points:

4 SIGMUND KOCH

1. Both group studies are "collaborative" but only in the special
sense that many creative men agreed to pursue, within the climate of
the Study, individual tasks of vital interest to themselves. The study is
not collaborative or "groupish" in any sense implying an intention to
relinquish individuality or even idiosyncrasy to some prissy conception of
the common weal.

2. The Study aims for no grand "integration" of knowledge. If a
"Summa Psychologica" or even a "Synopticon" had seemed even re-
motely within reach, our inclination would have been to abstain on
principle. The Study seeks to reflect the diversity of thought that actually
exists and is premised on th^jv^

proachjas^^ It conceives of its con-

tributor groups as pluralities of creative individuals who view those areas
which they know best through the screen of their own expert sensibilities.
The discussion outlines invite the play of individual sensibilities on com-
mensurable themes, thus helping the reader to collate positions.

3. The Study's aims are neither legislative nor evaluative. It rejects
all monolithic codes for the generation and processing of knowledge, or
for virtuous scientific conduct. If there is a central bias, it is for the
loosening of those constraints which can keep men from significant prob-
lems or thoughts through fear of the unorthodox.

The Study by no means devalues the insights of recent "science
of science" but would wish them set in a perspective better adjusted
to a field barely beginning to test established methods of science on an
inimitably diverse and intricate collection of subject matters. Such a
perspective can emerge by seeing the end-products of science as every-
where conditioned by human decision, value, creative option ; by freeing
from staleness that truism which holds scientific inquiry to be con-
tinuous with other human activities. Analysis in science, then, becomes^
more than a succession of routine tasks in the "logic" of science; it be-
comes an enterprise which can uncover the significance of its objects
only by holding in view the relations between creative process and
sentential product, strategic gamble and cognitive outcome. *"*

The grounds for the inclusion of the eleven formulations contained
in this volume are best conveyed by reference to the planning for the
total Study (cf. Introduction to Study I, Vol. 1, especiaUy pp. 21-27).
Eleven topics is a stingy allotment relative to the range and density of
effort in studies of the person and the social setting. If many significant
lines of work have been omitted, the ones included are also significant—
and sufficiently varied to suggest a generous range of the problems and
tasks that systernatists face.

Of the contributions relevant to "personality," three— Murray's

Introduction to Volume 3 5

"scaffold," Rapaport's systematic examination of psychoanalysis, and
Rogers's codification of his client-centered framework — are representative
of conceptual frameworks having broad scientific objectives and long-
standing influence. Cattell draws together certain proposals towards
systematic thought suggested by the logic and findings of one of the
principal methods of personality research (factor analysis). By reviewing
a delimited but basic problem area (psychogenetic studies of twins),
Kallmann gives an exhibit of the type of painstaking, stepwise empirical
work on which the resolution of issues common to many systematists
must depend.

Turning to the contributions that most would allocate to "social
psychology," again we find represented three lines of work which point
towards general systematic objectives. Each of these, moreover, stresses
a different one of three principal "levels53 at which social "variables"
may be conceived. Asch could be said to represent the type of social
psychology which sees no basis for conceptual separation from individual
psychology. Thelen and Parsons consider instances of the approaches of
"group dynamics" and "unified social science theory," respectively. An
important methodic formulation (latent structure analysis) is offered by
Lazarsfeld in response to a class of problems which he sees as ubiquitous
to psychology and social science. Finally, the contributions of Newcomb
and of Katz and Stotland present preliminary systematic assessments of
problems stemming from one of the more active fields of social psy-
chological research — the study of attitudes.

On one point, agreement among authors is so vehement that it
may be not unfair to anticipate it here. Personologists and psychologists
called "social" refuse in this volume to compress their concerns into the
standard compartments. If they have always seen their concerns as funda-
mental to the task of psychology, they now seem eager to assert this even
within the hearing of "fundamental" psychologists. To such refreshing
truculence we could but yield by avoiding the time-worn substantives
"personality" and "social psychology" in the title of this volume.

Mention of certain editorial provisions is in order. Readers will find
the complete statement of discussion themes, as sent to contributors,
reproduced in the Appendix. There is, of course, variation in the extent
to which the different presentations adhere to the discussion themes.
As an aid to readers interested in the detection of key convergences and
divergences of positions, index numbers corresponding to the principal
thematic items have been inserted, where relevant, in the individual
tables of contents appearing before each paper. The system of index
numbers is explained in the Appendix.

This final volume of Study I contains a section of general comment
on the study. An editorial epilogue presents a few trends suggested by

6 SIGMUND KOCH

the essays in all three volumes. Attention is restricted to conspicuous
trends which can give a "fix" on the position of systematic psychology
relative to its recent history. There is also a special supplement by Conrad
Mueller on certain methodological implications of the contributions in
sensory psychology. Dr. Mueller generously served in a capacity much
like that of sub-editor in the sensory area.

Further discussion of trends will be offered in the final volume of
the series, Psychology and the Human Agent. There is, however, no
standard "theory53 of the meaning of this study. If there were, we
would consider the main aim compromised. That aim has been to
develop materials of unique comprehensiveness and depth in terms of
which each reader may enrich his own view of systematic psychology.
The thirty-four essays of Study I can reward efforts towards secondary
analysis and synthesis — whether by student, specialized scholar, or gen-
eral reader — for a long time to come. Let there be as many theories of
this study as there are readers.

Psychology: A Study of a Science is the result of a project sponsored
by the American Psychological Association and subsidized by the Na-
tional Science Foundation. The project was known as "Project A" of the
"APA Study of the Status and Development of Psychology." The work
profited from the counsel of an Advisory Committee consisting of
Dael Wolfle, Chairman, and Clarence H. Graham, Lyle H. Lanier,
Robert B. MacLeod, Eliot H. Rodnick, M. Brewster Smith, and Robert
L. Thorndike. Howard H. Kendler, Lyle H. Lanier, Conrad G.
Mueller, and Karl E. Zener composed a Panel of Consultants to the
Director. The generous part played by the members of both groups is
described in the introductory sections of Volume 1.

PREPARATIONS FOR THE SCAFFOLD
OF A COMPREHENSIVE SYSTEM

HENRY A. MURRAY
Harvard University

Introduction 7

Interest in Significant Human Feelings, Thoughts, and Actions. Influence of

Medicine 9

Interest in the Earliest and Innermost Origins of Things 12

Interest in Process, Change, Differential Development, Creativity. Influence of

Chemical Embryology 14

Observations of the Interdependence and Hierarchical Integration of Functional

Processes: Adoption of the Organismic Concept. Influence of L. J. Henderson 17

Interest in the Directionalities and Effects of Overt Behaviors 19

Influence of Whitehead and Lewin: Concepts of Physical Field, Cathexis, Pro-
ceeding, Serial, etc 21

Additional concepts for the present scaffold 29

Cathexis 29

Dyadic system 30

Thema 31

Thematic dispositions 34

Serials 34

Ordination 35

Influence of Freud, Jung, and Other Psychoanalysts 36

Influence of Darwin, Bergson, and Other Evolutionists: Adoption of the Con-
cept of Creativity 38

Influence of Social Evolutionists, Cultural Anthropologists, and Sociologists . 45

Compelling Need for Comprehensiveness 47

Tolerance of Uncertainty 49

Interest in Systems 50

Apologia 52

References 53

INTRODUCTION

It seems that the majority of my voices are in favor of this enterprise,
for here I am, pen in hand, intending to comply so far as possible with
the editorial suggestions.

7

8 HENRY A. MURRAY

But a minority of me — and now surely, at the outset, is the moment
to give vent to it — believes that certain of the analyses invited by the
discussion outline are premature, not for all psychologists perhaps, but
for those who are concerned with human lives and personalities. The
topics suggested for discursive treatment are broadly defined; but,
even when taken with a grain of salt, the task calls for meticulous
criticism of one's own speech, semantic niceties, overelegant definitions.
Should not criticism and refinement be in balance with spontaneity, ex-
ploration, and invention if a science is to grow in a way and at a pace
appropriate to its age? Also, do we have sufficient data or sufficient
organization of the data to arrive at anything more than a miniature
system for a tiny region of transactions? Systematic psychology, being
very young, has occupied only a small portion of its legitimate terrain.
Its contemporary schools are like our thirteen colonies along the Atlantic
coast line, a narrow strip of provincial culture. Their manifest destiny
is to move West, order the wilderness with the best available tools, crude
as they now are, and eventually achieve a more refined and compre-
hensive system which embraces all parts and functions of the whole, the
total personality. At this stage I should hate to see our center of gravity
move any further to the side of perfectionistic rituals, a hair-splitting fussy
Conscience.

No doubt this large endeavor will bear fruit; but despite its promise,
it is not applauded at this moment by some members of my household
because of their suspicion that it is liable to seduce some promising
psychologists away from the study of personalities — the domain that is
theirs, and only theirs, to explore, survey, and map — away from the
humanistically important riddles which we should be creeping up on
gradually and craftily.

Another reason for my hesitation in joining this enterprise is the
impossibility of my adhering to the suggested ordinance of discourse. It
is evident that certain of its terms could be met only by psychologists
with other aims than mine. It is an admirable mold — straightedged and
nicely shaped — for exclusively experimental specialists, observers of
closely restricted animal activities, peripheralists, and positivists; but
literal adherence at all points is scarcely possible for naturalists, gen-
eralists, and centralists, who study gradual transformations of the dis-
positions, beliefs, and modes of action of human beings as they manifest
themselves in different social settings.

Despite the above reservations, twenty months ago it was decided
somehow that I accept the challenge as an adventure in self-discipline;
and, in conformity with the committee's outline, I went ahead with
what amounts to an intellectual autobiography in so far as this relates
to the development of my present scaffold for a theory of personality.

Preparations for Scaffold of a Comprehensive System 9

This part of the assignment was easier than I anticipated; but the second
part — setting up a logically articulated skeleton of the whole — was so
much more difficult that, despite an extension of time as well as every
possible guidance and encouragement from a most charitable Director, I
was unable to arrive at a satisfactory set of basic propositions before
the date line. In short, I proved unequal to the set standard. It happened,
however, that more than half of the matter to be ordered in Part 2 has
been included in Part 1, and so, the Director, pressed by generosity,
decided that the peculiar fragment which lies before you might serve as a
kind of substitute contribution. Its title might be this: certain orienting
dispositions, impressive observations, and influential theories as deter-
minants of scientific aims, assumptions, methods, and conceptions.

INTEREST IN SIGNIFICANT HUMAN FEELINGS, THOUGHTS,
AND ACTIONS. INFLUENCE OF MEDICINE

It is generally assumed by the uninformed and innocent that all
psychologists must have at least one "orienting attitude55 in common: a
stout affection for human beings coupled with a consuming interest in
their emotions and evaluations, their imaginations and beliefs, their
purposes and plans, their endeavors, failures, and achievements. But
this assumption, it appears, is not correct. A psychologist who has been
constantly prodded and goaded by these propulsions, as I have been,
belongs to a once small and feeble, though now expanding and more
capable, minority. Anyhow, this bent of empathy and curiosity toward
all profound experiences of individual men and women should be set
down as one of the prime determinants of several definitive decisions,
which shall be mentioned, respecting the scope of my scientific concern
and of a methodology to fit. This is a crucial point because, if my interest
in events of this sort had been less steadfast, I might have turned to more
manageable phenomena.

My interest in people, their doings and their ills, must have had
something to do with my choice at college of history as field of con-
centration and of medicine as career for later life.

The study of history implanted the idea of the time dimension as an
essential part of the very definition of reality as well as a miscellany of
coarse facts to support my speculations when I dipped, three decades
later, in the stream of sociology and anthropology. But the study of
medicine was more influential: it led to two years of surgery and five
years of research in physiology and in the chemistry of embryology, with
a Ph.D. from Cambridge University in physiological chemistry.

The practice of medicine taught me a lot of commonsensical things,
one of which was that among the few almost indispensable methods of

10 HENRY A. MURRAY

arriving at valid diagnoses (apperceptions, inferences) is that of inquiry
— the thorough detailed recording of the patient's memories of interior
sensations and pertinent emotional experiences. We were taught to
distinguish perceptible physical signs (overt sense data) and imper-
ceptible symptoms (reports of covert psychic processes) and to value
both. The proof obtained on the operating table, time and time again,
that a correct diagnosis of an abdominal condition could be made
solely on the basis of a patient's reported symptoms was so firmly im-
printed on the entablatures of my cortex that when, in later years, I was
confronted by Watson's dogma — his radical repudiation of subjective ex-
periences as material for psychology — my head assigned it to the category
of eccentric foibles. I was an empirical behaviorist, born, bred, and
trained, in the sense that every physicist, chemist, and biologist is neces-
sarily a behaviorist. But when it came to dealing with human beings, I
could see no advantage in allowing myself to be converted into an
exclusive, half-paralyzed behaviorist who, on metaphysical grounds, elects
to deny himself an invaluable source of data. (This does not apply to the
current cultural situation: today, after a complete semantic somersault,
every psychological process — perception, emotion, dreaming — is called
"behavior.")

My above-mentioned interest in people was not at all confined to
their physical activities — say, to the routes they chose and the muscles
they used in locomoting to the restaurants they preferred to ingest the
food that was most appealing to their senses. I was much more interested
in their feelings, evaluations, and conceptions relative to other matters,
and for the most part, so were they — and so were my militantly be-
havioristic friends of later years — more interested in the valued products
of their intellections than in their own muscular accomplishments. In
due couree, assured that correctness of prediction is the best index of the
relative worth of different methods, I did a few impromptu experiments
and found empirically that the most dependable single operation I
could perform in attempting to foretell what a behaviorist would do next
or in the near future was to ask him. But the commonsensical avowal I
wish to make here is this: that first as a doctor and second as a psy-
chologist I have never ceased to elicit direct expressions and reports of
interior experiences — somatic, emotional, and intellectual — not only as
sources of indications of overt actions to be executed in the future, but
as indications of occurrences that are intrinsically important. For ex-
ample, the occurrence of anxiety, or the persistence of unhappiness,
or the generation of a new theory is as important to me when taken as a
dependent variable (something to be predicted) as it is when taken as
a hypothetical or intervening variable (an aid in the prediction of some-
thing else). Though imperceptible to us and therefore inferential., covert

Preparations for Scaffold of a Comprehensive System 1 1

mental processes and products, some with and some without the property
of consciousness, happen to be intrinsically attractive to a cogitator of
my persuasion, and I see no insuperable barrier to their being incor-
porated in a unified body of scientific facts and propositions. If the
heavenly bodies had memories reaching back to the Big Bang and words
intelligible to us, what astronomer would shut his ears to them on
principle? Anyhow, this concern of mine, this reliance on a multiplicity
of inferences, checked and rechecked, this vision of a theoretical system
largely composed of psychological, rather than physical, variables, makes
it necessary for me to leave exclusive positivism to those who deal with
entities that are incapable of supplying us with valuable verbal repre-
sentations of what has occurred and is occurring behind their surfaces.
But enough said; I must return to my surgical internship and finish
listing what I learned that influenced subsequent decisions respecting
procedures and objectives in the field of psychology.

From medical practice I derived the "multiform method" of assess-
ment, coupled with the belief that it should be possible for a group of
trained collaborators using a wide variety of methods to make a reason-
ably complete examination, formulation, and appraisal of a whole person
as an ongoing order of differentiated functional activities. This objective
is achieved over and over again on the physiological level by practitioners
of medicine but when transferred to the psychological level its attain-
ment is impeded by innumerable special difficulties. To cut down my
hope to size — to make it congruent with what can feasibly be undertaken
under existing conditions, with available personnel, with existing con-
cepts and existing methods — has been my enduring but never sufficiently
successful resolution.

Also derived from medicine were consequential convictions respecting

(1) the determining importance of biochemical occurrences — say di-
gestion, assimilation, metabolism, excretion — after the organism has
finished eating and the interest of the average psychologist has faded;

(2) the ultimate scientific value of systematic, thorough, and detailed
case histories; and (3) the necessity of an adequate classification of the
entities and processes within the domain of one's elected discipline.

I have nothing more to say under the present heading except to
avow that my special interest in the dispositions and thoughts (rather
than the bodies) of human beings was one determinant of the rather
sudden decision I made to shift from physiology to psychology. Also
influential in some degree were the impressions ( 1 ) that human person-
ality, because of its present sorry state, had become the problem of our
time — a hive of conflicts, lonely, half-hollow, half-faithless, half-lost, half-
neurotic, half-delinquent, not equal to the problems that confronted it,
not very far from proving itself an evolutionary failure; (2) that psycho-

12 HENRY A. MURRAY

analysis had already made appreciable progress in exposing and interpret-
ing the deeper processions of the mind; and (3) that my temperament
was more suited to the making of coarse maps of newly explored areas
than to the refinement of relatively precise maps of familiar ground.

INTEREST IN THE EARLIEST AND INNERMOST
ORIGINS OF THINGS

It seems that I was scarcely four years old when, like a cornerstone,
the law was laid in me that storytellers should begin at the beginning.
The beginning was not only engaging in itself, but necessary to an
appreciation of the rest — all succeeding adventures, stratagems, con-
flicts, loves, and triumphs of the hero. I felt with Aristotle : no beginning,
no excitement at the climax, no catharsis. And so, if my father or my
mother failed to start a fairy tale with "In the beginning/' or its
equivalent, "Once upon a time," I knew that I was about to be deprived
of essential information and this, in my book of rules and regulations,
was ground for protest.

But more consequential than this early requirement for a good fairy
tale was my first down-to-earth attempt to latch on to the beginning of
a course of actual events. The attempt was prompted by a sudden bellow
that originated, I soon discovered, from a strange baby in my parents'
room. Puzzled, I was told that this noisy creature was my brother and
perfectly adorable. Here surely was a notable beginner; but what was
the beginning of this beginner? My inquiry ended with the answer that
Dr. Anderton, my mother's red-bearded physician, had brought him in
his bag, the very bag from which I had so often seen him lift spatula,
swabs, and stethoscope.

That I should have rested — I won't say comfortably — with the words
"doctor's bag," that I should have abandoned my quest for basic knowl-
edge after one essay, not followed the path of my intent, the path of
infinite regressions, one leg further at the least, a step which would have
taken me to the place where Dr. Anderton obtained the babe, that I
should have quit so soon, is evidence of a docility or squeamishness so
unsuited to the career of science that even now I blush to acknowledge
it in print. If all along I have been stopped at the very verge of the
unknown by some constitutional timidity, it is possible, yes probable, that
I have failed to see, or to interpret properly, or to report candidly
occurrences that were beyond the stretch of well-established scientific
theories or beyond the bounds of embedded moral sentiments.

I have mentioned my halt at the "doctor's bag" conception of the
fount of life — suggesting parenthetically that I might not have Iain down
too happily with this solution — I have mentioned this defeat of curiosity
as a possible indirect determinant of what eventually became a con-

Preparations for Scaffold of a Comprehensive System 13

firmed interest in the earliest and innermost origins of things. This hypoth-
esis might help to explain why, twenty-five years later, I elected to
spend the greater part of three years in an incubator with several dozen
eggs, observing and measuring the chick embryo's earliest manifestations
of vitality. The point is that I managed at long last to get inside the
doctor's bag, or, better still, at 103.6°F, almost within the womb of the
beginning of a beginner. Peering through a microscope, through a little
fabricated window in the egg's shell, spellbound as any libidinous
voyeur, I witnessed the procession of momentous transformations that
mark the hours when the embryo is no bigger than an angel perching
on a pin point. Here, it seemed, were occurrences of great significance
which into concepts no contemporary intelligence could digest.

The same hypothesis might serve, in some measure, to account for
my disappointment, if not aversion, when I encountered the science of
psychology at college and listened for a while to what was considered
worth announcing about the perceptual processes of the adult mind, the
mind of a Western intellectual, a mind without a history, strapped to a
piece of apparatus in the laboratory. Also in keeping with this hypoth-
esis was my subsequent embrace of Freud with all his facts and
legends respecting the earliest months and years of life. Freud kept my
first commandment : he began at the beginning. In my initial enthusiasm
I hardly noticed that he never reached the consummation of the allegory,
the heroic adult and his tragic end.

Depth psychology was obviously my meat. In thedepths one came
upon the earliest and most determining dispositions. Whatever initial
doubts I had respecting unconscious psychic processes were soon enough
dispelled. Several weeks with Dr. Jung at different times, three years
with Dr. Morton Prince, an orthodox Freudian psychoanalysis, and a
period of training with Dr. Franz Alexander and Dr. Hanns Sachs, ten
years of therapeutic practice — these experiences were hugely influential in
shaping my personality and my thought. But at no time, to the annoyance
of my friends, was I a good Jungian, a good Freudian, a good Adlerian,
or a good schoolman of any breed. I held all my teachers in high esteem,
but judged that each of them — necessarily at this stage of theoretical
development — was more or less one-sided. The notion which invited me
was that of attempting, with the aid of additional ideas culled from the
writings of McDougall, of Lewin, and of my colleagues at Harvard, a
preliminary revision and integration of current academic and psycho-
analytic theories to accord with a large collection of reasonably solid
facts obtained by the multiform method of assessment. This effort re-
suited in the crude blueprint for a system which a number of us sub-
mitted in Explorations in Personality [3], a blueprint which stressed the
earliest and least accessible determinants of behavior. We did not do
this to the satisfaction of the psychoanalysts, first, because all behavior

14 HENRY A. MURRAY

was not traced to infantile sexuality and aggression and second, because
we classified overt actions as they occurred, whether or not we had
reasons to believe that they were subsidiary to deeper, hidden aims.

But now, if I may, I shall mention another disgrace of childhood
which seems relevant to this topic — what I have called my interest in
origins and 'beginnings. If, awhile back, I almost disqualified my child-
hood self as a potential truth seeker, by mentioning that moment of
scarcely pardonable poltroonery in the face of the Great Riddle, what
I now have to confess is evidence of something bordering on complete
damnation in the scales of science. Not going beyond Dr. Anderton and
his bag signalized a defect in daring and determination to solve prob-
lems; but worse than this is the inability to know a pithy problem when
you see it. So far as I can recall, if truth will out, I was never prompted
to ask about the very beginning, the beginning of mankind or the
beginning of the world. Passively and without suspicion or comment I
received the news that some six thousand years ago God — who, in
pictures I was shown, had a big beard, not red like Dr. Anderton's, but
white as my venerable and remote grandfather's — that God had con-
structed the first man in a single day, and, a little later, molded from one
of this man's ribs the first woman, et cetera, et cetera.

I suspect that it was the swallowing and digesting of this fable, trust-
1 fully and without complaint, which determined, to some degree, my
gust for Darwin and the evolutionists who succeeded him, as well as the
joy I felt in shedding the constraining creeds of orthodox religion. It
was as though a strait jacket had been unfastened and I stepped out to
breathe and move and think for the first time without embarrassment.

It was from biology and chemistry that I received the exciting notion
that man is descended from the very humblest of parents, a more or
less fortuitous combination of chemical elements — such low-caste stuff
as hydrogen, oxygen, carbon, and nitrogen — and that, instead of a day,
it took two billion years or more to shape him. Also noteworthy was the
evidence that the wondrous evolutions of man and of his productions
may be credited, in some measure, to the very tendency which in the
Garden of Eden version led to his disgrace and fall, that is, the inborn
tendency to explore and to experiment among forbidden things.

My enthusiasm for this theory becomes more intelligible when
viewed in conjunction with the next orienting disposition to be listed.

INTEREST IN PROCESS, CHANGE, DIFFERENTIAL DEVELOPMENT,
CREATIVITY. INFLUENCE OF CHEMICAL EMBRYOLOGY

It is hard to decide: should I speak here of a predisposition that
sensitized me to a certain class of facts or should I speak of a certain

Preparations for Scaffold of a Comprehensive System 15

class of facts which engendered a disposition to accept them and look
for more of the same kind? I have always thought it good emotional
policy not only to enjoy, so far as possible, the inevitable, but to will
the obligatory. In this case, however, I am inclined to stress the inner
bent ahead of the compelling facts because to the majority of psycho-
logical theorists these facts are not particularly compelling.

I am referring to facts which particulary attracted me during my
studies at the Rockefeller Institute of the physiological ontogeny of
chicken embryos. To summarize a long story, what seemed both most
obvious and most important about the interior of the embryo were (a)
the givenness, the inherent spontaneity, of its cellular activities and (b)
the continuous sequence of orderly metamorphoses (clearly perceptible
under the microscope) which resulted from these activities, and hence
the necessity of including formative (constructive) processes in one's
scheme of variables.

Unintrusive observation was enough to nail down the self-evident
proposition that chemical and physical activity, metabolism and move-
ment, are integral properties of every animate body, things to be in-
cluded in the very definition of life. Also, it appeared that organic
processes are not only primarily endogenous, autonomous, and proactive
(initiated and sustained from within, rather than being merely reactive
to external stimuli) but especially in the early stages of development
are, so far as one can see, not perfectly coordinated with other processes,
not constantly directed toward the achievement of effects extrinsic to
themselves. This view of things was initially implanted by observing,
time after time, the very first beat of the uncompleted embryonic heart
and noting that it contracted irregularly and then regularly for quite
a while — I forget how long precisely — before the blood vessels and the
corpuscles were far enough along in their development to make it
possible for this organ to perform its predestined function, namely, to
pump oxygen-refreshed blood through the arteries of the body. The
primitive heart was merely exhibiting its emergent capacity to contract,
like a playful child or puppy, achieving for a period no effects outside
its own growth of form and potency.

This notion of endogenous, initially undirected and uncoordinated,
process-activity constrained me, in later years, on the one hand, to
qualify my acceptance of the fashionable stimulus-response formula,
with its implicit assumption of a nothing-but-reactive organism and, on
the other hand, to qualify my acceptance of the proposition that all
activities are motivated. In short, I take "life" — say, the ceaseless
processes of metabolism — as given, just as Newton took motion as given,
and do not look for something antecedent to it, except in an evolutionary
sense.

16 HENRY A. MURRAY

The other influential impression I received from my studies of
embryonic physiology was that, during the first phases of its career, a
relatively large proportion of the totality of processes within a living
organism is involved in the development of somatic substance, in the
work, let us say, of anabolism, of incorporating and combining new
elements, and so of constructing and of reconstructing parts unexampled
in the history of that particular unity of animation. In other words, the
most significant characteristic of the embryo is not so much the arrange-
ment of its perceptible component forms of matter at this or that
moment, as its activity in forming and transforming forms of matter.
Defining "energy53 as the capacity to produce change, change of relations,
we can say that most of the energy of the embryo is devoted to generative
changes, that a host of processes precede forms, one of the effects, or
"functions," of some processes being to build and to rebuild them. That
is, the organism, being an open system (as Bertalanffy showed me later),
selects from its environment, incorporates and synthesizes, potentially
energic matter, and thereby increases its resources, taking a course
opposite to that defined by the second law of thermodynamics (which
applies to closed systems). Furthermore, clearly perceptible under the
microscope were divisions of the soma into regions and in each region
the production of distinctive structures, in short, morphological dif-
ferentiation, preparatory to specialization of functions.

It was these observations of embryonic developments, besides what I
could understand about the science of energetics, which initially pre-
disposed me to stress "mythologies53 of energy, process, change, function,
more than "mythologies55 of matter, structure, permanence, and to regard
the organism as ordered successions of different kinds of processes, the
effects of some of them being primarily internal — formations and re-
formations of component structures — with a re-ordering of the processes
occurring consequentially. In short, according to this way of thinking,
creativity is an inherent property of the organism and stability is another.

Four of the ultimate resultants of my interest in process, develop-
ment, and creativity were ( 1 ) the adoption of the whole history of an
organism, the entire life span of a personality, as the macro-temporal
unit that requires formulation (although it may be half a century before
a satisfactory way of doing this — an adequate conceptual scheme and
an adequate methodology — is devised) ; (2) an interest in all manifesta-
tions of significant changes of personalities — progressive transformations,
eliminations, and reconstructions, learnings, extinctions, and relearnings,
regressions and deteriorations — and in the determinants of such changes,
and hence a special, but by no means exclusive, emphasis upon the in-
fluential experiences of childhood; (3) a devotion to all forms of the
imagination — dreams, fantasies, prospections, ordinations (plans), plays,

Preparations for Scaffold of a Comprehensive System 17

story-constructions, myths, rituals, religious conceptions, works of art,
and scientific speculations — as manifestations of involuntary and largely
unconscious process-activities which, when influenced by a strong and
continuing intention, may, in some cases, have a definitely creative out-
come; and (4) the construction of a large number of methods (most
of them unpublished) for the eduction and exposure of imaginal proc-
esses and products (so-called projective tests) .

Imaginal processes and products appeal to me not only because of
their intrinsic interest, but because they have been shown to be the best
source of dependable clues of underlying (and often unconscious) dis-
positions and conflicts of dispositions. An often verified hypothesis is
that some of these inferable dispositions are residua of deformative in-
fantile experiences and that a few of them are prodromes of conditions
in the offing.

OBSERVATIONS OF THE INTERDEPENDENCE AND
HIERARCHICAL INTEGRATION OF FUNCTIONAL PROCESSES:
ADOPTION OF THE ORGANISMIC CONCEPT. INFLUENCE

OF L. J. HENDERSON

It was in 1920, during my studies of oscillations of the physicochem-
ical equilibria in the blood, under the tutelage of Lawrence J. Hender-
son, author of The Order of Nature [1], that I first became familiar
with the organismic, or organismal, proposition, as formulated by E. S.
Russell in 1916 and elaborated by W. E. Ritter in 1919. Belief in its
essential validity was confirmed a little later (sometime before I heard
anything about gestalt psychology) by observations of the embryo — per-
ceiving the sequential effects of Spemann's genetical "organizers" — and
by studies of the regulatory functions of the autonomic nervous system
in conjunction with the endocrines. Clearly demonstrable in higher
animals are vertical integrations of superordinate and subordinate loci
of control, levels of directors and coordinators, "lines of command"
starting from some center in the segmented neuraxis, or lower brain,
and ending in regional plexuses and local nervous networks, a hier-
archical system, depending on "feedbacks" (as we say today), which
executes the genetically determined "domestic policy" of the organism.

Here it might be appropriate to refer to Cannon's principle of
homeostasis, and to the fact that consideration of the radical develop-
ments during the embryonic period led me to stress the concept of
progressive disequilibrium, continuity through expansive, constructive
change, as a supplement to that of homeostasis (which is more applicable
to the interior activities of adult organisms). The concept of homeostasis
(the maintenance and, if disturbed, the restoration of the same state

18 HENRY A. MURRAY

of equilibrium) is a basic scientific induction, defining as it does the
measurable relationships of multifarious interdependent elements and
processes, relationships which either persist unchanged, or, if modified
by some intruding element of exigency, are in due course reestablished.
But it should be noted that this principle, as commonly defined, is valid
only within a certain time span. The time span varies with the age of
the organism as well as with the system (physiological, psychological, or
sociological) that is under consideration. In the body of a healthy adult,
the morphological, physicochemical, and physiological relationships are
quite stable, or soon restabilized, over a period of many years despite
the slow changes which eventually result in the signs and symptoms of
senescence. But in the embryo homeostasis has virtually no span at
all, or an extremely short one: the organism as a system being char-
acterized in all its manifestations by perpetually changing states of
equilibria, states that move in an irreversible direction. In short, the
embryo is in disequilibrium or, at most, transitional equilibrium from first
to last. Comparable, I thought later, though less striking to the eye and
less susceptible to precise measurement, are the seasons of transitional
equilibrium on the psychological level, which occur most obviously in
childhood but also later, during the early phases of some new enterprise,
let us say, or when the creative imagination is steadily advancing. At
such times psychological processes are transformative, and when they
terminate, the person is a different person, or his sphere of relationships
is different, and there is a different equilibrium to be sustained.

Although I came away from my embryological studies with a firm
belief in the unity of the organism through change, in orderly dif-
ferentiations and integrations, my medical training had established a
special vigilance in respect to signs and symptoms of functional imper-
fection, and I soon discovered how normally abundant are such evidences
on the psychological level, evidences of disunity, of retardation, deviation,
deformation, and retrogression. It appears that millions of years of
evolution have resulted, on the one hand, in an almost perfect system,
let us say, of somatic operations, and on the other hand, in a human
brain which contains at birth no comparably ordered system of depend-
able proactions and reactions, but instead, a matrix of potentialities in
a relatively amorphous state, potentialities for unprecedented develop-
ments of talent, at one extreme, and for idiocy and lunacy, at the other.
Hence, especially for human beings, life is a continuous procession of
explorations, surmises, hunches, guesses, and experiments, failures and
successes, of learnings and relearnings — aging consisting of a sequence
of gradual and occasionally abrupt indurations (rigidifications, solidifica-
tions, fixations, hardenings), both of forms and functions. Consequently,
a psychologist has to deal conceptually with doubt, distrust, indecision.

Preparations for Scaffold of a Comprehensive System 19

and postponement of behavior among his subjects, together with oc-
currences and continuities of competition and paralyzing conflict be-
tween their dispositions.

My bent toward organismic, holistic, molar, or "gestalt" con-
ceptions of the personality and its activities scarcely fitted me to wax
avid when I encountered, later on, the then dominant elementalistic,
connectionistic, chained-reflex, molecular theories of learning, theories
that were being hungrily ingested by all who cared on what side their
academic bread was buttered. Of the two fallacies, reductive and seduc-
tive, so nicely discriminated by Herbert Feigl, I was more liable to the
second, though I had no use for those lazy white elephants of the mind —
huge, catchall, global concepts signifying nothing. Eventually I was
persuaded by Professor Boring — more generous of his time than any
teacher I ever had — that the principles of elementalism and associa-
tionism are applicable under many circumstances, especially, let us say,
to the establishment of certain neurotic symptoms as well as to condi-
tionings that occur below the level of conscious control or when the
mind is tired or confused and functioning below par. In short, elemen-
talism (emphasis on parts, integrants, components) and holism (em-
phasis on wholes, integrates, ordinations of components) are necessary
complements.

INTEREST IN THE DIRECTIONALITIES AND EFFECTS
OF OVERT BEHAVIORS

One passes by inseparable gradations from an interest in the auto-
nomic-endocrinal coordination of the multifarious somatic processes of
the body and in the local effects of their different operations, to an
interest in the cortical ordination of sensory, muscular, and verbal
processes toward successive achievements of different overt effects, most
of which endeavors, if successful, contributing in some way or other to
the well-being of the total organism.

Hence, it was already in the cards I held that, on entering the do-
main of psychology, I should very soon become concerned, not so much
with reflexes and patterns of muscular movements, as with the various
changes effected by such movements and the changes in the states and
thoughts of other people effected by spoken words and sentences.

The fundamental fact, it seemed to me, is the survival of the living
organism, the continuation of its metabolic processes, and the dependence
of this procession upon the periodic attainment of a number of distinct
effects, such as the inspiration of oxygen, the expiration of carbon
dioxide, the ingestion of water and food, and the excretion of waste
products. The different processes, modes, and subeffects whereby the

20 HENRY A. MURRAY

same kind of terminal effect is achievable in different species of organisms
or even in the same organism at different times were, at the start, a
matter of considerable, but subsidiary, importance. A man shivering
with cold may restore optimal body temperature by moving to a warm
place, putting on an overcoat, closing doors and windows, lighting a
fire, turning on the radiator, taking a hot drink, or exercising strenuously.
Defined in terms of physical vectors (locomotions or manipulations in
space) these are different actions, but the beneficent effect in all cases is
the same. Indeed, a person may obtain all necessary "goods" with the
minimum of activity on his part: they may be furnished providentially
by nature, gratuitously by parents or friends, or in exchange for money,
by domestic servants or employees. I had a good friend who lay in
bed, blind and completely immobilized from his neck down, for twenty
years. He had a sensitive and brilliant mind that was bubbling over
with unimpaired effectiveness and charm until his death, and yet he saw
nothing and never moved a muscle. Every act necessary to his survival,
to the stimulation of his feelings, and to the increase of his knowledge
had to be performed by someone else. This was but one of countless ob-
servations which persuaded me of the necessity of providing concepts for
the analytical dissection, whenever necessary, of any short segment of
activity into ( 1 ) kinds of exciting initial situations, ( 2 ) kinds of processes
(e.g., covert psychic processes, overt psychomotor or psychoverbal
processes) with or without kinds of utilities (e.g., tools, weapons, con-
veyances, telephone, typewriter, etc.), (3) kinds of modes, or styles, of
processional activity, kinds of psycho-expressive processes (e.g., speed,
grace, gestures or tone of voice expressive of uncertainty, anxiety, self-
confidence, anger, good will, deference, compassion, etc.), and (4) kind
of effect (change from the initial exciting to the terminal gratifying
situation).

In my persistent efforts to move, step by step, toward an adequate
solution to such problems, I was greatly assisted by the reported observa-
tions and formulations (1) of biologists from Darwin on, and of
others, particularly McDougall, who had used the concept of instinct
as their tool; (2) of Freud relative to the sex instinct, aggression, and
anxiety, and of Adler relative to the craving for superiority; (3) of
Tolman and other animal psychologists who had carried forward the
endeavor to define and measure rigorously different drives; (4) of Lewin
with his constructs of tension system and of quasi need; and (5) of
sociologists regarding the wants of men for status and for power.

In Explorations of Personality I attempted to define a number of
actional dispositions which, in the absence of a less objectionable desig-
nation, were termed "needs" (or "drives"). These constructs proved use-
ful in categorizing inferentially the overt behaviors of the subjects we

Preparations for Scaffold of a Comprehensive System 21

studied as well as the behaviors of the characters in the stories they com-
posed. But this particular working inventory of human drives (kinds of
motivations, purposes, intended effects, goals) was, in several respects,
deficient, and ever since, these deficiencies have kept provoking me to
prolonged efforts to conceive of fitting remedies. An account of today's
resultant of my arduous and still continuing endeavor to arrive at a
more comprehensive and integrated system will be presented in a later
work.

Before leaving this topic I should say that I have not been satisfied
to limit my objective to the formulation of overt behaviors, certainly not
to the formulation of purely physical behaviors. Indeed, after perceiving
that the food-ingesting activities of animals and of men are not at all
representative of the majority of human actions (as Maslow has pointed
out) but, being most readily formulated in physical terms, are repeatedly
used nonetheless to illustrate this and that concept or generalization or
to serve as foundation for this and that postulational system, and that
they thus constitute an alluring conceptual trap for the unwary theorist
— perceiving all this, I established in myself a prohibition (which I
guiltily break occasionally) against using the hunger drive and its en-
suing motor patterns and effects as paradigm of directional behaviors or
even as a reliable reference point for speculation.

As I see it, a psychologist should be concerned not only with the
formulation of overt interpersonal verbal communications, the imme-
diate (intended) effects of which are changes of some kind among the
dispositions, evaluations, represented facts, interpretations, or commit-
ments of the other person, but also with the formulation of covert intra-
verted mental activities, the immediate (intended) effects of which are
such things as: a better interpretation and explanation of some recalled
event or of some current physical symptom, a reevaluation of one's own
enactions (past behaviors) or present abilities, the definition of the con-
tent and boundaries of a required concept, the composition of the plot
of a story to be written, the resolution of a conflict between two purposes,
or the ordination of a plan of action (tactics) to be executed at some
future date.

INFLUENCE OF WHITEHEAD AND LEWIN: CONCEPTS OF
PHYSICAL FIELD, CATHEXIS, PROCEEDING, SERIAL, ETC.

I owe much to the incomparable Alfred North Whitehead and the
incomparable Kurt Lewin, nothing less than the conviction that con-
crete reality is to be found only in the momentary. With theoretical
physics in mind, Lewin devoted a good deal of his unusual imaginative
powers to the definition of space constructs, topological and hodological,

22

HENRY A. MURRAY

the momentary field; whereas Whitehead, founding his penetrating re-
flections on organic and mental phenomena, emphasized the momentary
process^ the perpetual becoming and perishing of "actual occasions" and
the historic continuity or progression of these occasions. Although I have
never gained sufficient understanding of Whitehead's terminology to
apply the categorial scheme of his philosophy of organism to the realm
of ordinary human experience and behavior, I am indebted to him for a
number of conceptions which I have revised to suit the purposes of a
psychologist. First among these is the concept of an event, or fact, as a
participation of processes in which two or more interdependent entities
are involved occurring in a certain place or along a certain path, within
a certain medium, through a certain segment of time, and resulting in
a certain kind of change. I conceive of a range of events of different
molarities. Theoretically, an ultimate submicro event would have the
smallest spatial scope (smallest containing field), the smallest entity
scope (fewest component particles), smallest process scope (fewest dis-
tinguishable changes), and shortest temporal span (duration). (For ex-
ample, it is estimated that tau and theta mesons are composed and de-
composed in about a hundred-millionth of a second. ) Some micro events
occur within the boundaries of solids, i.e. entities that can be treated as
solids under most conditions (anything from a crystal to a planet), but
others are integrated, synchronously and sequentially in time, in such a
way as to constitute an event of greater scope and span, and this event,
in turn, can be seen to constitute a necessary part, or phase, of an event
of still greater scope and span, something that can be defined in terms
of a single resultant process — secretion of a hormone by one cell, a single
color sensation, influxion of a single image, contraction of a single mus-
cle fiber — or in terms of a longer or more massive process — secretion of
saliva, contraction of the heart, perception of a configuration, momen-
tary feeling, evaluation of an object, movement of a limb, etc. Such an
event may be a part of a yet larger, longer whole — say, a stimulus-
response unit (perception, apperception, and evaluation of a pertinent
entity, concurrent emotion, actuation of a pattern of muscular move-
ments against resistance, production of an effect, perception, appercep-
tion, and evaluation of this effect) . Thus, by increasing step by step one's
scope and span of concern, one arrives at the largest and longest defin-
able unit of activity, a macro event. A personologist usually has to deal
with macro events, or proceedings; and from the fullness of each of these
he abstracts those variables which are relevant to his purpose, in the
knowledge that numberless other variables will be unrecorded and hence
omitted from his formulation. Thus, the major concepts of the scaffold
to be built — such as need, entity, configuration, process, succession,
effect, place, route, time— are all considered to be abstractions from an
event or progression of events.

Preparations for Scaffold of a Comprehensive System 23

As twentieth century inhabitants of the Western world, we seem to
be living and acting — partly as a consequence of our acquired Indo-
European language — in euclidean space, moving about on the support-
ing surface of an assumptively permanent material planet amid a great
variety of substantial objects, inanimate and animate, natural and arti-
ficial (man-made), some transient, some relatively permanent, each with
its distinguishing physical attributes. "Such presumptions," as Whitehead
says, "are imperative in experience," and "in despite of criticism," we
still employ them "for the regulation of our lives." And so, for better or
for worse, I too have employed them, not only in the regulation of my
life but, with certain qualifications, in the regulation of my theorizings.

If I were forced to choose one side of the age-old antinomy between
the "metaphysics of substance" and the "metaphysics of flux," my tem-
perament would decide in favor of the latter, the version of the universe
that is linked in our minds with some vivid sentences attributed to Hera-
clitus. But, as I see things and events, it is not necessary to go to one
side or the other, either of this classical division between different aspects
of nature or of other dichotomies, such as that between matter in space
and motion in time, or between instantaneous configurations of material
bodies and modifications of these configurations, or between chemical
structure and chemical properties and processes, or between form and
function, or between anatomy and physiology, or between entity and
activity, or between actor and action, or between noun and verb. It is
possible to choose both sides and combine them in single propositions.

Perhaps ray most influential basic model is that of biochemical
metabolism, repetitive and restorative as well as progressively and irre-
versibly transformative: the lifelong succession of compositions, decom-
positions, and re compositions of concrescences and perishings, of vital
chemical substances. Here is incessant flux certainly, with the catabolism
of anabolized materials liberating the energy for every manifestation of
vitality (thermal, chemical, electrical, mechanical — emotional, disposi-
tional, mental, and muscular) ; and here also are countless instantaneous
configurations of substances within cells, of cells within organs, and of
organs within a body, some parts of which (skeleton, ligaments, con-
nective tissue, skin) are relatively solid and enduring like the framework
of a house. Consideration of anabolisms, in which two or more chemical
entities combine to form or to re-form a more complex entity, where
one can attribute the course of events to no single actor and his act, has
led me to conceptualize, in many cases, systems of participant entities
and participating processes rather than placing the major burden of de-
termination on one person or on one person's conscious purpose. Here
one might think of the mental participations involved in creative activity,
with conscious intention playing but a minor role, or of the emotional,
verbal, and actional participations of two lovers.

24 HENRY A. MURRAY

At this point let me explain for clarity's sake that in view of the
mind's tendency to "spatialize" everything, as Bergson pointed out, and
in view of the ambiguous usages in the social sciences of such words as
structure, configuration, form, pattern, integration, etc., I prefer to re-
strict the word configuration to the instantaneous (transient), and the
word structure to the enduring, spatial relations of the substantial com-
ponents of an entity, assemblage of entities, or region (extended surface
area), and to use the word succession to designate the once-occurring,
and the word integration to designate the recurrent temporal relations
of the component processes of a proceeding (uninterrupted activity,
endeavor, interaction). According to this terminology it would be
proper to speak of the structure of a house, of a painting, of an organ-
ism, of a chemical compound, of a crystal, or of an atom; and it would
be proper to speak of the integration of mechanical, electrical, chemical,
mental, verbal, or musical processes, through a certain period of time.
One could also speak, in a highly abstract way, of the hypothetical
structure of the mind or of the personality, although mind and person-
ality are known to us only through successions of covert (subjective)
and overt (objective) processes. You see I am wary of the word "struc-
ture," because, if used to describe concatenations of activities one gets
that impression of permanence, regularity, and lawfulness which is so
dear to the hearts of scientists and yet so incongruent with the facts in
many instances.

The debt I owe to Lewin can be most simply set forth if we restrict
thought, for the time being, to the motor activities of one person from
the beginning to the end of a single simple proceeding, or endeavor, a
goal-directed and goal-attaining course of action. In such a case the
"whole" effect (attained goal) of the pattern of muscular processes can
be defined by designating the relevant differences between the structure of
the physical field at the initiation of the activity and the structure of the
physical field at its termination. This will tell us what the person "did" —
he moved, let us say roughly, from one location to another, or moved
an object from a table to his mouth, or put a new tire on his car, or hung
a picture over his desk, etc. But more than this, ideally considered —
and here is where Lewin comes in — a sufficient characterization of the
field at the start of the activity, and at every instant from then on, would
set forth the immediate determinants of the over-all direction of the
activity as well as of each successive part, or unit, of the whole. As
Lewin put it, "the behavior b at the time t is a function of the situation
S at the time t only," where S denotes the total situation (field) — the
field of forces within the person (internal situation) as well as the field
of forces exterior to the person (external situation), as apperceived and
evaluated by the S, The initiating total field (a momentary cross section,

Preparations for Scaffold of a Comprehensive System 25

or time-slice, through everything that is influential) determines the be-
havior resulting in the next field, which, in turn, determines the behavior
resulting in the subsequent field, and so on, until the occurrence of an
act resulting in a field which determines the cessation of that variety of
endeavor. In Lewin's scheme of constructs, as in mine, the major variable
of the internal situation (internal field) is some kind of excitation (with
direction and magnitude) — a quasi need, need-aim, or drive; but here I
am stressing the external situation.

It should be noted in passing that an adequate formulation of the
immediate, or antecedent, determinants of behavior can never be given
in terms of the instantaneous external situation (configuration of space
or of objects or of forces in space). Even in the extreme case of a wholly
stationary external environment one must take account of the process
through time of the subject's perception, apperception, and evaluation
of the situation; and this brings us to Whitehead's actual occasion, the
"real thing." In most cases, the so-called momentary external situation
(set of antecedent determinants) is likely to consist, not so much of a
spatial configuration, as of a rather long pattern of symbolic processes,
such as a paragraph of instructions read to the subject by an experi-
menter. But, let us return to the simple case of a stationary physical
field in which a mobile person is positioned and consider what kind of
map should be made of this so-called momentary situation.

Man being a terrestrial organism for the most part — for the duration
of this discussion, anyhow — the space to be represented will be a two-
dimensional flat surface, natural or artificial — either a circumscribed
area of ground (composed of rock, soil, or sand) or a floor area within
a building. This area we shall call the territory (the total spatial scope
of our concern), and this territory (say, a sparsely settled rural area)
we shall divide into regions., and these regions into subregions, and so
on indefinitely, if necessary, until we arrive at a multiplicity of places.
Each region will have a certain area and shape and will be distinguish-
able from other regions by the number, position, and physical attributes
(size, shape, color, etc.) of its occupants (say, an assemblage of trees, of
potato plants, of weeds, or of buildings), or by the absence of occupants,
and/or by boundaries (walls, fences, hedges), not to speak of brooks
and rivers. Furthermore, there will be strips with smooth surfaces con-
necting some of the differentiated regions, which I shall call routes, one
of which will run through a subregion occupied by buildings, each
marked by sets of symbols, one set indicating that food may be purchased
there, another indicating tools, another drugs, and another clothes. Let
this suffice as an account of the structure of the space relevant to our
problem. Now Lewin was shrewd enough to see that a map of such a
territory showing the location of physical objects and their attributes^

26 HENRY A. MURRAY

mere patterns of sense data, or mere primitive perceptions, is of little
relevance to psychology. A modern artist, by a conscious effort, might
view his environment in this way, or possibly a visitor from Mars; but
even in the above-given bare description of the territory I could not
without misunderstanding omit such words as trees, potato plants,
buildings, fences, routes, food, tools, drugs, and clothes, all of which
words refer in a rough way to objects which not only make themselves
known to our senses by means of their physical attributes, but which,
under certain conditions, are capable of contributing to (or, in other
cases, subtracting from) our welfare. Hence, it is not so much the
physical attributes as such but the known or supposed man-pertinent
capacities of objects which influence behavior (including the capacity
of some objects to delight the aesthetic sensibilities of the subject) . It
was these pertinent capacities that Koffka and Lewin had in mind when
they spoke of the "behavioral," or "psychological" environment, the
environment of meanings or significations.

This point of view was congenial to the one at which I had arrived
with the help of Uexkiill. Accustomed to the distinction between the
attributes and properties of chemical compounds, I had made a com-
parable distinction between what a human object "looks like" and what
he "does" under specified conditions. Here I am leaving out, for the
moment, what an alter does to the subject solely by virtue of her or his
physical attributes (beauty, ugliness). What an alter does, the kind of
thing he does, to the subject., I called a press (plural: press}. For ex-
ample, the press of Mr. X vis-a-vis a given subject might be "to animate
him (the subject) intellectually," just as the usual press of the drug
Benzedrine when taken by mouth is "to stimulate mental processes." The
capacity to stimulate is one of the biochemical properties (latent press)
of Benzedrine, and when Benzedrine passes into the blood stream the
property becomes manifest as a process distinguished by its effect.
Similarly, a known alter, regarded from the subject's point of view, can
be represented as an assemblage of subject-pertinent properties, or latent
press, which will be manifested as processional effects (operative press)
either spontaneously or after appropriate stimulation, when the subject
and the alter meet. Thus, as I saw it, the physical structure of the en-
vironment was representable in terms of the geometric configuration
of regions, places, and objects, each with its potentially effective subject-
pertinent properties (latent press). That strip of smooth surface over
there is called a route because from position A to position B it has the
property of supporting a human body or a conveyance and of facilitating
locomotion; and boards cut from those trees have properties suitable for
the excluding walls and supporting floors of houses, etc., etc. As a con-
sequence of countless past experiences, such properties seem to be re-

Preparations for Scaffold of a Comprehensive System 27

vealed to us Immediately by mere perception, but at this point I prefer
to speak of apperception, or apperceptive perception, since it is con-
venient and often important to distinguish verbally betweeil the clear
impression and identification of a particular kind of object — say, a
hickory tree — and the realization of its properties — say, the properties
of hickory which make the wood especially suitable for ax handles. The
chief difference between the conceptualization of a pertinent property
of an inanimate object, such as a drug, and a pertinent property of a
person (alter) is that, in the case of the latter, one is dealing with a
mobile object whose activity may be unprovoked by the subject, and
one must distinguish between an endeavor that fails (through incapacity)
and an endeavor that succeeds. It is the difference between a pressive
disposition without ability and a pressive disposition with ability. But
more of this later; I must return to my topic.

There was a wide gap, it always seemed to me, between Lewin's
symbolic constructs on the level of physics (representative of public
physical events) and his constant references to a miscellany of wholly
private psychic processes in his subjects which he cleverly distinguished
by intuition, but which he spoke about as if they were overt and obvious
to everyone, or could be reliably inferred on the basis of observed be-
haviors. Not many psychologists realized so clearly as did Egon Brunswik,
that for Lewin the exterior field (the environment) was within the sub-
ject's head. What Lewin called the "psychological environment" is the
subject's apperceptions of the environment — a necessary construct; but
it stood alone, no place having been provided for a more "objective"
definition of the environment, say, as apperceived by the psychologist, by
selected judges, or by the conventional majority. Thus by Lewin's
scheme it is not possible to distinguish between a morbid delusion and a
realistic, or congruent, estimation of the external situation: the situation
is exactly what the subject thinks it is, or more accurately — since
Lewin rarely, if ever, asked a subject — it is what you think the subject
thinks it is as you empathically perspect his thoughts during the course
of his behavior. Furthermore, if the humanly pertinent properties of other
environmental objects (as estimated by the psychologist) are never
mentioned, we shall never know how much of the external situation was
rejected by the subject.

As a step toward the clarification of this issue, a number of us,
stimulated by an extended definition of Freud's important concept of
projection, conducted numerous investigations of differences between the
external situation as carefully and systematically perceived and apper-
ceived, say, by a consensus of trained observers (the alpha situation),
and the same situation as perceived and apperceived (under conditions
less favorable to accuracy) by subjects with different personalities (each

28 HENRY A. MURRAY

a beta situation), or by subjects in a certain experimentally engendered
temporary state. This is the sphere of concern which is now called "per-
sonality and perception/'1

The bulk of our experimental findings were unanimous in their
verdict respecting the importance of dispositions (interests, evaluations,
and needful tensions) in determining the outcome of perceptual, apper-
ceptual, conceptual, compositional, and ordinational (planning) pro-
cesses. In short, as antecedent determinants of overt behavior, one must
include, not only the structure of properties and processes of the con-
fronting exterior situations as arranged by the experimenter (cluster
of independent variables, or alpha situation), but perceptions and
apperceptions of certain of these things (beta situation) as determined
by the dispositional state of a given personality or type of personality
(cluster of intervening, hypothetical, or conventional variables) .

Besides many other things, this meant to me (with my memories of
chemistry) that a psychologist will bring in less knowledge by viewing
a person as a mass-point of indifferent constitution in a field of forces,
as Lewin (with his interest in physics and his image of Galileo at the
tower) was tempted to do, than he will by viewing him as an entity
with a particular conjunction of distinguishable properties. It is, of
course, true that in establishing some sorts of lawful relationships be-

*In this and in other related enterprises, fortune favored me with early col-
leagues of the stature of Erik H. Erikson, Donald W. MacKinnon, Saul Rosen-
zweig, R. Nevitt Sanford, and Robert W. White, of William G. Barrett, Kenneth
Diven, Isabella V. Kendig, Walter C. Langer, Christiana D. Morgan, and Carl E.
Smith; later of Thelma G. Alper, Leo Bellak, Vera V. French, Elliott Jaques,
Robert R. Holt, Daniel Horn, Morris I. Stein, Silvan Tomkins, and Frederick
Wyatt; and, more recently, of Gardner Lindzey, of Anthony Davids, Richard V.
McCann, and Robert N. Wilson. I have also been advantaged by collaborations,
all too short, with Freed Bales, Tamara Dembo, Cora DuBois, Walter Dyk,
Jerome D. Frank, Christopher Fried, Asa Koht, Philip Lichtenberg, Goodhue
Livingston, Charles C. McArthur, H. Scudder McKeel, James G. Miller, Merrill
Moore, Hobart Mowrer, Benjamin J. Murawski, and Henry W. Riecken, as well
as with a host of others on the OSS assessment staff during the war years, of
whom Edward Tolman and John Gardner have, in interior dialogues, admonished
me most often. Among these warm friends and coworkers I have no reliable way
of apportioning the credit for leading me to relatively valid concepts and for
canceling many of my least propitious errant speculations, and no reliable way
of apportioning the blame for withholding criticism at moments when I might
have been deterred from this or that cognitive folly. Anyhow, I am grateful for
the opportunities I have had to serve as one of many channels for the ebullient
ideas that have swirled and eddied round the table at the Harvard Psychological
Clinic. And here I must make public my profound indebtedness to my good
friend and critic, Gordon W. Allport, staunch champion of minorities, without
whose timely advocacy the Clinic might have been dissolved and left no wrack
behind.

Preparations for Scaffold of a Comprehensive System 29

tween entities it is possible to disregard differences of constitution, but
even in physics, how often can one predict the outcome of an experi-
ment without taking into account the internal structure of the molecules,
or such properties of substances as conductivity or melting point? In
short, on the down-to-earth empirical level (as contrasted with the
sphere of transcendent, or purely hypothetical, entities) one must include
in one's formulations the properties (in specified states) of the entities
engaged in the observed transaction. For example, some material entities
are nourishing to human organisms, others stimulating, others soporific,
and others lethal, and one property of some human organisms In a
certain (suicidal) state is to select a lethal rather than a nutritive
entity for incorporation. I would delete these references to the obvious,
if it were not for the fact that most of us, in our endeavors to be ob-
jective, to formulate behavior in terms of perceptible movements — say
approach and ingestion for survival — forget, for example, that poison
is attractive to persons in a certain state. In short, we cannot throw
Aristotle to the dogs and restrict our diet to the more elegant formulas of
Galileo: chemistry is still among the reputable sciences and closer to
psychology — think of oxygen, digestion, metabolism, and endocrines —
than is its more admired older brother.

Another related conclusion supported by our findings was that the
historic succession of the dispositions and experiences of a scientist has
a great deal to do with the concepts and theories that he comes out with,
and largely because of this conviction, I have often taken pains, as by
request I am taking now, to expose my inborn and acquired bents and
biases, rather than to make a great to-do about my exemplary scientific
objectivity. It happens that one of my inductions from experience is
that many of those who spend most type asserting their immaculate
empiricism are somewhat below average in their awareness of the
distorting operation of their own preferences and ambitions and, there-
fore, are more liable than others to sally forth with reductively incon-
gruent versions of reality.

Additional concepts for the present scaffold. Among the other
conceptual consequences of our studies of personalities and their apper-
ceptions of other personalities and of my attempts to analyze single
proceedings, six may merit definition.

Cathexis, From Freud I gratefully accepted the concept of cathexis
(value, valence) as a useful variable in formulating personalities as well
as single interactions of personalities. But instead of limiting its applica-
tion to a loved person (the power of an alter to attract, enchant, and
bind the affections of a person), I defined it as a possible disposition-
evoking capacity of any kind of entity, or of any kind of activity of an
entity, chiefly the capacity (1) to excite attention (interest, concern.

30 HENRY A. MURRAY

thought, talk), or (2) to excite attention plus evaluation, either positive
(favorable — say, gust, wonder, admiration, love, approval) or negative
(unfavorable — say, disgust, contempt, disapproval, distrust, resentment,
fear), or (3) to excite attention plus evaluation plus pertinent activity.
All types of entities seem to be capable of such evocation — a certain
kind of food, a homestead, a utility, a person, a social institution, a
novel, a moral code, a scientific theory, a philosophy of life — and
similarly capable are all types of activities of entities. Not only a total
entity, but any part, integrant, or component activity of an entity may
have the power to attract attention, to please or to displease, to instigate
activity. You may like a person as a whole but not like certain things
he does, or you may like certain things he does but dislike him as a
whole. A father spanks the boy he loves because he hates lying and hopes
to spank this habit out of him, and so forth. The same might be said
of the negatively cathected (and hence deleted) parts of a book in proc-
ess of composition, a book which, taken as a whole, may be highly
cathected by its author.

My present notion of cathexis is not far from the elaborate definitions
of it that were published in The Clinical Study of Sentiments [4], except
now the more favored term is "value" and the concept has been in-
corporated in a larger system. The term "sentiment," "attitude," or
"established evaluation" points to dispositional property of a personality
which corresponds to the cathexis of an entity. One can say that subject
A has a strong sentiment or attitude (established disposition) pro X, or
that his consistent evaluation of X is highly positive, or that X has a
high positive cathexis or value for A. Both terminologies are useful. The
concept of cathexis is also useful, perhaps most useful, in indicating the
subject's effect on other people: in what quarters and to what extent
he will evoke positive evaluations, based, say, on affection, erotic love,
admiration, or compassion, and leading to accessions or invitations, as-
sociations and conjugations, compliances, services, or donations, etc.,
in what quarters and to what extent he will evoke negative evaluations,
engendered by disgust, contempt, moral condemnation, or envious
resentment, and leading to rejections, exclusions, decessions, expulsions,
or inflictions, etc. It is not sufficiently acknowledged, I surmise, that a
full characterization of a personality should include, as does the char-
acterization of a chemical compound, the varieties of dispositional effects
the subject has on different kinds of alters.

Dyadic system. The notion came and stuck that a dyadic (two-
person) relationship, whether transient or enduring, should be formulated
as a single system, equal analytic attention being devoted to each
participant. Although I have never been inclined to accept Harry Stack
Sullivan's restriction of the domain of psychology to the sphere of inter-

Preparations for Scaffold of a Comprehensive System 31

personal relations, I use dyadic interactions as a test of every formulation
or theoretical system I encounter in the literature. If the proposed set
of antecedent environmental variables does not provide for the definition
of an alter's subject-oriented verbal or physical behavior (e.g., such kinds
of "stimulation" from the alter as petition or command, praise or repri-
mand, inquiry or offering of information, expression of good will, and so
forth), if it does not provide tools of this sort, then the system is not
suited to the representation of the great bulk of human reactions. It may,
of course, have other virtues, but not those I require: variables appro-
priate to the prediction of concrete social episodes.

Thema. The idea matured that the basic pattern of a single dyadic
interaction might be most simply represented by 2, a symbol denoting the
immediate direction, the need-generated orientation (goal), of the pro-
activity emanating from the first interactor, followed by z'z, a symbol
denoting the emotional response of the second interactor, and when in-
dicated, a symbol denoting the need-generated orientation (goal) of his
reactivity. Whether the goal of the first interactor's (subject's) activity
is the aim of an independent need (and hence intrinsically satisfying if
achieved), or the aim of a subneed (satisfying if achieved although it is
no more than a subordinate component of a large system of need-aims),
or the aim of a quasi need (merely instrumental and hence not in-
trinsically satisfying) would be a question for further investigation.
Months of antecedent study and subsequent exploration might be re-
quired to determine the probable status, or relative potency, of all the
needs involved in a single sentence. The same applies to the need-
determined response of the alter. On this level of formulation (the
formulation of a single proceeding), it would be sufficient to represent
the immediate need-aim of the subject (proactor) and the need-response
of the alter f reactor) . The need-response of the reactor, viewed from the
subject's stamlpoint, has been termed a press, the alpha press being the
alter's actual response and orientation (in so far as he and the psy-
chologist can define it) and the beta press being the subject's apper-
ception of the alter's response and orientation. The simplest formula,
then, would be either an N-P (if the subject initiated the interaction) or
a P-N (if the alter acted first). This I termed a simple micro thema, a
simple macro thema being an over-all, and hence much coarser, formula-
tion of a longer transaction, and a serial thema being an articulated
procession of simple micro themas, which might or might not be
representable as a macro thema.

I might clarify this a bit by illustrating complementation, the simplest
type of dyadic thema (others being reciprocation, cooperation, competi-
tion, opposition). Let us assume two interactors: X a confirmed trans-
mitt or and Y a confirmed receptor; and then, out of a large number of

32 HENRY A. MURRAY

complementary needs, let us choose the following pairs, and finally, let
us assume that in each case the episode is completed to the satisfaction
of both parties (criterion of a veritable complementation) .

Subject X, transmitter Subject F, receptor

Need to inform (to relate facts, rumors) Need for information (state of interest,

curiosity)

Need to explain (to interpret events) Need for explanation (state of perplexity)

Need to counsel (to give advice) Need for counsel (state of indecision)

Need to amuse (to tell a funny story) Need for amusement (readiness for mirth)

Need to console (to express sympathy) Need for consolation (state of distress)

These pairs can be taken to represent the state of affairs in a dyadic
system, at the start of five different proceedings. The transmitter is
characterized by the tension of a valued fullness (pleni-tension] — he has
a mental possession and the need to impart it — whereas the receptor is
characterized by lack-tension, that is, by a need for something, something
which, in this case, the other person is capable of giving him. Assume, for
example, X has a mental possession (a funny story) which he is keen to
communicate and Y is keen to hear a funny story. As a rule, there will be
mutual satisfaction if the story strikes Y as funny and he responds with a
hearty laugh. Further analysis may reveal that the apparently pleni-
tensive transmitter has nothing very interesting to say but merely a
strong (processional) disposition to babble (verbosity), and/or a lack-
tensive need for attention and appreciation. Similar is the next type of
dyadic pattern, reciprocation, except in this case we have a reciprocal
complementarity, the second phase being marked by a reversal of roles —
the former receptor transmits with an appropriate degree of zest and the
former transmitter receives with due appreciation.

Consideration of long sequences of interpersonal themas of this sort
has pretty much confused me respecting the proper usage of the venerable
S-R concept. The intended effect (need-aim) of much proactive talk
(reactive to the mere sight of another person) is an appropriate kind
of sympathic response (press) from the alter (e.g., expression of agree-
ment, compliance, interest, mirth, affection, admiration, gratitude, and
so forth), and there seem to be a good many hypomanic (chemically
stimulated) self-starters and transmitters in the world who, instead of
predominantly responding to other persons, sail forth each day full-
freighted with, a miscellany of impatient stimulations for any ac-
quaintance (releasor) who might be capable of the complementary
responses; and when a conversation is once launched, every response
is a stimulus to a response which is a stimulus to a further response, and
so forth, until the tidy S-R model has been so thoroughly rolled through
all things that it looks as if it needed treatment, some sort of radical

Preparations for Scaffold of a Comprehensive System 33

rehabilitation. Perhaps it has already been rehabilitated, without my
knowledge, by the more advanced S-R theorists.

It became evident in due course that a simple thema, whether micro
or macro, is no more than a very coarse, though often meaningful and
convenient, classification of an episode. To formulate an episode in a
more refined way numerous other variables must be included until one's
initially simple representation of its major dynamic components has been
transformed into a complex thema. Among the immediate determinants,
for example, of Y's positive or negative reaction to a "funny story" told
by X, might be the "appropriateness" of the situation (never mind now
how this is judged), the relative status and degree of intimacy of X
and Y, the mirth-potency of the story, whether it is new or stale to Y,
how well it is told by X, whether Y is momentarily at odds with X,
the current mood or state of Y, the acuteness of Y's sense of humor in
general and for this kind of story in particular, how fastidious is his
standard of wit, to what extent is Y's system of values susceptible to
offense by this kind of story, and so forth. Just as some psychologists
have profitably devoted a professional lifetime to the study of a hungry
animal in a maze containing food, so might others spend rewarding years
in investigating the interior and exterior determinants of any one of a
hundred other common types of themas, say, a thema with an un-
successful or unexpected outcome, such as "the joke that fails," "the
command that is defied," "the conjugal proposal that is rejected," "the
injury that is forgiven," and in each case, why?

My own attempts to practice what I am now preaching — to explain
in some detail the course of a single type of interaction — have been
spotty and rather crude, and, for the most part, this side of publication.
Christopher Fried, Philip Lichtenberg, and I have separately spent two
years or more investigating a few of the determinants of the dyadic
themas that occur during film-recorded competitive and cooperate at-
tempts to arrive at a common plan of action; and, of course, there have
been countless "clinical" occasions for minute perceptions of other
common patterns. But, on the whole, the facts compel me to acknowl-
edge that, except for resolute endeavors over the last twenty years to
analyze and formulate the apperceptible processes and products that
occur during impromptu compositions of dramatic stories, I have not
focused long enough on any single type of thema or on any single method
of observation and measurement to come out at last with a brilliant
cluster of decisive findings. Decision has been difficult, because if a
would-be personologist should elect to devote his energies to the building
of a miniature system of postulates and theorems applicable to the under-
standing of one kind of thematic unit, he would have no time for the
observation of other varieties of behavior; hence he would never get

34 HENRY A. MURRAY

around to the practice of his profession, namely, the investigation of the
interrelations of the more determining gross components of personality.

Thematic dispositions. It has become more and more apparent to me
that the energic components of personality can be better defined as
thematic dispositions than as general actional dispositions. For example,
instead of saying that X possesses the trait of aggressivity, or that he has
a ready and intense need for aggression, one should, if possible, specify
the nature of the pertinent press (stimulus) and say with more precision
that two of the properties of his personality (I won't translate this into
symbolic shorthand) are supersensitive dispositions to react with re-
sentment and aggressive words (1) to apperceived insults to his self-
respect and (2) to apperceived vainglorious boastings by an alter.

Serials. I was slow to perceive that current psychological theories of
behavior were almost wholly concerned with actions of relatively short
duration, reflexes and consecutive instrumental acts which reach their
terminus within one experimental session, rather than with long-range
enterprises which take weeks, months, or years of effort to complete.
Here, it seemed, was one of the most striking differences between men
and animals, namely, the capacity for time-binding (Korzybski) or the
span of time-perspective (Frank, Lewin). The behavior of animals can
be explained so largely by reference to attractive or repellent presenta-
tions in their immediate environment and/or to momentarily urgent and
rather quickly reducible states of tension; whereas a great deal of a
man's behavior cannot be explained except by reference to persistent
"self-stimulation" in accordance with a plan of action, which often in-
volves the subject's commitment to a distal goal or set of goals, as well
as to a more or less flexible (or rigid) temporal order (schedule) or sub-
sidiary, or stage, goals. Observing his behavior over several months or
years, we see, not only the recurrence of a large number of patterns
devoted to the repetition of valued experiences and the prevention of
disvalued experiences, patterns with homeostatic effects, but a number
of interrupted successions of proceedings (which I am calling serials,
or long enterprises), each temporal segment of which is progressively
related to the last (carrying on from where the other stopped), though
separated from it by an interval of time (commonly a day) . A successful
serial is different from many day-by-day reactions in so far as its effects
are transtatic rather than homeostatic, that is to say, it transforms or
transcends the existing steady state by carrying a person from one level
or form of equilibrium — dispositional, material, ideational, or social — to
another: a new interpersonal relationship (an additional commitment)
becomes established; a new house is purchased and furnished (which
must hereafter be kept up) ; knowledge is gradually assimilated, and a
new orientation (directing one's efforts toward another target) is

Preparations for Scaffold of a Comprehensive System 35

acquired; the subject graduates from college, gets a job, and takes on the
responsibilities of a new office; a novel is written and published, and
so forth. Progressive enterprises of this sort constitute the bulk of a
healthy young adult's endeavors in a "civilized" society.

Ordination. It took me years to realize that the psychology of the1
higher mental processes had been equivalent, in the niinds of most
psychologists, to the psychology of cognition, and that the psychology of
cognition was largely concerned with the processes whereby a person
acquires objective knowledge and understanding of his physical environ-
ment— the very processes and the very aims which are dominant in us
psychologists — and that i, the more fundamental and important knowl-
edge of the satisfying and dissatisfying, the beneficial and the harmful
properties of the environment and of the self s capacity to cope with
them, and ii, the still "higher" mental processes involved in the con-
struction of a plan of action, were pretty generally neglected. What
should we call the persistent, self-critical, conceptual, and often logical
mental processes that continue over several months in the mind of a
psychologist until they terminate with the construction of an integrated
design for his next experiment? These processes commonly take off from
perceptions and explanations of previous experiments and results; but
their immediate aim is not so much to conceptualize already observed
events (cognition), but to imagine something unobserved — new condi-
tions and new experimental operations — and, by logic or intuition, to
predict the outcome. During his months of planning the scientist (or
anyone else for that matter) is more frequently thinking, one might say,
on the efferent, rather than on the afferent, side of the cortical arc, and
some psychologists might, therefore, be disposed to subsume his mental
processes (processes which sometimes occur very rapidly — within a few
seconds) under conation, on the grounds that their function is to orient
and coordinate action. But against this is the fact that they are often very
"intellectual" (higher mental processes in the strictest sense), engaged
in a most difficult endeavor (since rational prediction is usually harder
than rational explanation), and superordinate to other processes, in the
sense that the goal and strategy which is ultimately selected will deter-
mine behavior for a good many months to come.

For better or for worse I have been calling such mental processes —
processes concerned with the selection and integration of plans of action
— ordination. The preliminary processes of the imagination — fantasies
and trial experiments in the mind — I am calling prospections. Here, in-
stead of entertaining recollections (replicative imaginations of past
events) , the subject is concerned with the future, prospectively picturing
himself in this and that situation, seeking this or that opportunity for
gratification or for the advancement of his ambitions. Here creativity

36 HENRY A. MURRAY

may operate to a marked degree. The selection from numerous alter-
natives of a concrete and specific goal, purpose, or aim to appease one
or more needful dispositions, I am calling orientation. It is the subse-
quent phase — the selection and temporal articulation of ways-means,
strategies, or tactics (represented by images or words) — that I am
calling ordination. I have found that the word can be used without
confusion, both for the process of constructing (ordinating) a plan and
for the construction (ordination) that results from this process. An
ordination may have a very short or very long time span; it may be
vague and global or clearly differentiated into discrete behavioral units;
it may be disjunctive or conjunctive (temporally integrated in a logical
manner) ; it may stand at any point along the rigidity-flexibility con-
tinuum; and it may have more or less of the property (power) of "im-
perativeness" (indicated, partly, by shame or guilt if adherence to the
ordination is imperfect) ; and so forth. One significance of this concept is
its discrimination of a major antecedent determinant of behavior in a
"civilized" society, namely, a fixed schedule, the time set for a certain
kind of activity, a prearranged appointment, a prescribed order of pro-
cedure— quite regardless of the mood, dispositional state, need, or what
not, existing at the moment. A good part of socialization consists in ac-
quiring the capacity to keep promises, and hence, to do something which,
at the appointed time, you are not inclined to do. Furthermore, we need
a concept of prospective time reaching into an imagined future, some of
which is filled (committed, planned) and some of which is still unfitted
(open, available for use) .

INFLUENCE OF FREUD, JUNG, AND OTHER PSYCHOANALYSTS

I came to psychology via Jung's Psychological Types and his Psy-
chology of the Unconscious, the first of which initiated my professional
interest in types of human nature, and the second, my interest in uncon-
scious processes as revealed by mythologies and religious imagery as well
as in the more central and integral transformations of personality. What
I gained from Freud was somewhat more specific and more applicable
in practice and, in due course, became so much a part of my regular
and irregular modes of thought that there have been times when I forgot
my debt and took his huge gift for granted. In the late twenties and
early thirities when Freud's name and works were anathema to the
majority of academic psychologists, I was a staunch advocate and de-
fender— as I am now — of his greatest contributions: (1) evidences of
the theory of unconscious psychic processes and their effects, (2)
evidences of the determining importance of early family relations and of
the experiences of childhood, of the persistence of complexes established

Preparations for Scaffold of a Comprehensive System 37

in those years, (3) countless illustrations of the multifarious manifesta-
tions of the sex drive, (4) division of the personality into id, ego, and
superego (conventional constructs), (5) definition of several mecha-
nisms— repression, isolation, denial, etc. — that operate in the service of
adjustment, of self-esteem, and of serenity of consciousness, and a host of
other more restricted constructs and theories illustrated by abundant case
material.

I was one of the founding members of the Boston Psychoanalytic
Society and throughout the thirties was so closely identified with its
cause that President Conant decided, primarily on these grounds, that
I was not qualified for tenure. Similarly, in the opinion of the psy-
chologists who reviewed it, Explorations in Personality [3] was a treatise
out of Freud, or, more accurately, an attempted adaptation of psycho-
analytic theory to academic standards. In short, what I have seized from
Freud is so very obvious that it should not be necessary for me, at this
late date, to lay it on the line.

The present situation is entirely different: Freud has conquered. He
has captured a large portion of the Western mind, his revolutionary
theories are learnedly and respectfully discussed in General Education
courses, he is now an indispensable fixture in the domain of psychology,
and so venerated by his professional disciples that his most casual com-
ments are repeated ritualistically as absolutes. Clearly his position is
assured and what we all owe to him is plain. The danger now is precisely
the opposite of what it was in the twenties when it looked as if professors
were built to shut their minds to him. Caught up as we are today in a
great wave of Freudiolatry we are inclined to take it all as gospel, to feel
that the greater part of what the Master said is so astute that the gestalt
which he created should not be spoilt by calling attention to a few trivial
defects. This attitude would have been impossible to Freud himself and
if continued its only consequence can be sclerosis of the mind and rigor
mortis.

As I weigh it, Freud's contribution to man's conceptualized knowl-
edge of himself is the greatest since the works of Aristotle; but that his
view of human nature is exceptionally — perhaps projectively and in-
evitably— one-sided, an extraordinary abstraction from the abundant
facts of life, facts which may have little bearing on the etiology of
neurotic symptoms but great relevance to other issues. My chief objection
is the commonplace that in his system, the libido has digested all the
needs contributing to self-preservation, self-regard, and self-advance-
ment, together with a host of others, and rebaptized them in the name of
Sex; and that sex itself is never given either its profound evolutionary
status or its interpersonally creative status. In the last analysis, it is
reduced to transient, superficial, localized sensations. But then, who

38 HENRY A. MURRAY

at this preliminary stage of knowledge can cover everything and be
right?

INFLUENCE OF DARWIN, BERGSON, AND OTHER EVOLUTION-
ISTS: ADOPTION OF THE CONCEPT OF CREATIVITY

My Heraclitean concern with process, change, and transformation,
dating from incubator years at the Rockefeller Institute, did not gain the
impetus of a possession until, with Lucretius vastly and vaguely in the
background, I came upon Bergson's theory of creative evolution, Lloyd
Morgan's concept of emergence, Whitehead's philosophy of organism,
Liebniz's monad, and the speculations of L. L. Whyte, Oparin, Wald,
and others, respecting biochemical evolutions. What I abstracted from
these authors, in conjunction with a few miscellaneous influxions from
the "unconscious," brought me to the conclusion that creativity — the
formation of new and consequential entities and of new and conse-
quential patterns of activity — is a centrally determining capacity of
nature, more especially of human nature. I had observed the progress of
morphological maturations in the embryo and later, the establishment of
new ordinations of serial enterprises and of new tactical patterns and
skills in personalities; but not until I paid attention to analogous pro-
ceedings on the physicochemical, sex-genetical, societal, and symbolic-
representational levels and in the sphere of technology, did I arrive at a
general conception of formative, or constructive, processes operating
throughout nature.

What does this amount to? First, a comprehensive generalization re-
specting a widely distributed capacity of entities, namely, under favor-
able conditions to associate and remain associated, to combine and re-
main combined, to become involved in the creation of new entities with
previously unexampled properties, and thereby to participate in the
making of an irreversible route of events. Finding manifestations of such
formative capacities at all integrative levels, we become more assured
of their importance, more convinced that they deserve a place in our
catalogue of fundamental dispositions. Also, we are invited by the pos-
sibility that detailed investigations of new productions at one level may
suggest analogies, correspondences, and hypotheses to be tested at
another. Second, the observation that matter has formative capacities
makes us realize that creativity is immanent in nature, not the pre-
rogative of some transcendent craftsman, such as Plato's Demiurge or
the Yahweh of Genesis, nor imposed on nature by the will of man. On
the one hand, it permits a natural explanation of some of the phenomena
on which the doctrine of vitalism once built its case, and, on the other
hand, it shows us why the term "mechanism" (with its implicit reference

Preparations for Scaffold of a Comprehensive System 39

to a man-made machine as model) was not the happiest choice to
characterize the procession of open systems under natural conditions.
Our conviction that the old vitaBst-mechanist opposition is a dead issue
is supported, I believe, by the abandonment of classical mechanics by
physicists and chemists as basis for their theoretical inventions. Third,
the addition of the formation (creation, construction, reconstruction)
process and effect to our inventory of dispositional properties of per-
sonality provides us with the otherwise missing necessary factor not only
for an adequate conception of the liveliest course of mental processes
through time (the work of the imagination), but for the systematic
representation of the functional interdependence of other members of
the inventory during the growing, expanding, and developing phases of
a person's life.

The concept of survival in one or another guise — self-preservation,
continuation, maintenance, homeostasis, and so forth — can fulfill the
same construct-integrating function in a theoretical system that is de-
signed to apply to mature lower organisms, since the great majority of
their activities may be partly understood historically, in terms of their
generally beneficent contributions to the continuation or the restoration
of a steady state. But the principle of survival is applicable only to the
status quo, not to mutation resulting in ontogenetic and phylogenetic
emergent evolutions. In my view of affairs, anyhow, it is necessary to
put the processes of composition and decomposition at the center of
things, between the terminus of the afferent side and the initiation of
the efferent side of the energy conversion arc of personality.

But this is not the accepted view today — despite our great concern
with learning, with developments of personality, and, very recently, with
some forms of mental creativity. The Freudian inventory of drives, for
example, includes sex, aggression (destruction), and anxiety-avoidance,
but not construction. Construction — which, being exemplified on the
chemical level, is more fundamental, in my view, than any of these
instincts as operationally defined by psychoanalysts — is subsumed, in a
vague and general way, under the concept of sublimation of infantile
eroticisms. Similarly in other special fields — sociology as well as ex-
perimental psychology. It may be a matter of time-perspective. If we
are in the habit of performing short experiments with a peripheral sub-
system of personality, no products of formative energies may strike our
apprehensive mass; but if we take a longer view we are struck by nothing
else.

Let us assume a comfortable position on Ganymede, satellite of
Jupiter, about two billion years ago and with supernatural eyes take a
morning look at the surface of this planet. We shall perspect, according
to those who are entitled to a guess, nothing save a fairly hot solution

40 HENRY A. MURRAY

of inorganic salts keeping company with the simplest carbon compounds
and enveloping this broth an atmosphere of gases from which oxygen is
absent. In the evening let us take another look. Since we have temporarily
assumed the power of a celestial being, a thousand ages in our sight
is as a day gone by and we shall now be in the twentieth century gazing,
I should hope with wonder, at a tremendous miscellany of natural
productions — 500,000 kinds of organic compounds, over 250,000 species
of plants, over 1,000,000 species of animals already identified by man.
We shall perceive numberless societal formations: human beings almost
everywhere, behaving rather regularly as members of a family, clan,
tribe, state, or nation, small or great, with fairly consistent governments,
laws, and policies. More obvious will be the territorial and habitational
constructions : land masses studded with settlements, villages, towns, and
cities, surrounded by cultivated fields and connected by paths, roads,
boulevards, and iron rails, running through tunnels and over bridges.
How long would be a catalogue of man's material manufactions,
architectural, mechanical, electrical! Think of the palaces and temples,
tools and armaments, machines and dynamos, waterworks, heating
systems, lighting systems, automobiles and airplanes, and gadgets by the
millions. Enough said. In the name of brevity, let's skip the rest and
consider the manifold combinations of sounds — the songs and symphonies
— and the combinations of images and imagined episodes — the mythol-
ogies and dramas, sonnets and heroic epics, histories and novels, and
their representations in paint, wood, and marble — and the combination
of concepts and reflections — the ethical philosophies, mathematical for-
mulations, and scientific systems — which engage the minds of men, and
with these let's end our swift survey of entities and activities on the
earth's skin. All these things, all varieties of social governments, material
conveyances and utilities, symbolisms and ideas, are productions of
the human part of nature, and in all probability, the vast majority of
them had their genesis in the imaginations of a single individual or of a
cluster of individuals.

And yet, the word "imagination" has been absent from the index
of most textbooks of psychology, and one has to search diligently to find
a little reference here and there to planning processes (prospection and
ordination), and despite the emergent interest in creativity, only a few
authors have seen fit to include, in some indefinite guise or other, a
formative disposition — habitational, implemental, interpersonal, social,
or symbolic — among the properties of human personality.

Darwin was primarily concerned with the occurrence of successively
more effective variations of mature morphologies from generation to
generation. In his day, biochemical science was not so far advanced as
to assist him with suggestions of plausible hypotheses respecting the

Preparations for Scaffold of a Comprehensive System 41

determination of these gross changes. Knowing nothing of the role of
chromosomes and genes, of nucleoproteins and DNA, it could hardly
be realized that chemistry is the instrument of heredity. Today, however,
we can reasonably postulate the creation of new genes along the route
of evolution, the mutation (by the transposition of a single atom within
a molecule) of a gene, and a stupendous variety of possible combina-
tions of genetical clusters from male and female. The chemists of
Darwin's time were not prepared to cope with the problem of the
emergence of living entities from nonliving entities, the virus was un-
known; and the physicists were speculating about matters other than the
possibility of the evolution of increasingly complex chemical elements
and compounds, say, out of light atomic nuclei. No one had yet suggested
that as the universe expanded new matter was constantly coming into
being. In psychology, prevalent interests and conceptions were far from
the idea that formative (gestalt-making) processes were involved in per-
ception and apperception, not to speak of their engagement in the
psychologist's own business of making concepts and formulating proposi-
tions. In short, the data necessary for a systematic representation of
constructive processes on different levels of integration were not avail-
able in the nineteenth century. Today, however, a multiplicity of facts
and of reflections are at hand, enough, it seems to me, for a rough
preliminary draft of meaningful analogies.

The very briefest outline I can devise, omitting several important
vectors and all details, includes the movement (motility, exploration),
and hence, by chance, the inevitable contiguity of different entities, one
or each of which is inherently attractive to the other — attraction
(gravitation, valence, cathexis) being one of the ever-present forces of
the universe — and, consequently, either symmetrical or asymmetrical
accession (approach) resulting in an association or structural formation
(creation, construction, synthesis, conjugation, or incorporation of a
smaller by a larger entity) new to this planet, and the cohesion, the
sticking and staying power, and hence the relative stability and longevity
of this unprecedented form of whatever category — organic compound,
genetical configuration, family relationship, tribal federation, govern-
mental law, religious belief, creed, or rite. If the established form is to
have further evolutionary value it must have the attribute of plasticity,
or flexibility, the capacity, that is, to play a part or to become involved
in subsequent transformations or reconstructions. The picture is one of
continuity through change. Only by losing its particular identity, by
perishing as such, can a variation become a link, stage, or episode, in an
evolutionary sequence, such as the one and only sequence that led to
the human species.

Some of my more earnest and literal-minded friends remind me that a

42 HENRY A. MURRAY

psychologist should abjure fantasies of temporal omniscience and keep off
of Ganymede. Formative processes lie outside the sphere of psychology :
they occur in the "depths," behind the scenes, take a long time to get
worked out, and are wholly unpredictable. A psychologist should attend
to the precise particulars of today's circumscribed field of observation.
Agreed, but suppose I ask one of these friendly critics to serve as a
subject and request him at the first session to demonstrate his ability
to design an experiment which will confirm or unconfirm a hypothesis
that is unfamiliar to him. In the second session he might be asked to
invent two different parables to illustrate the evil effects of fanaticism,
and in the third, to outline a course of action that might happily
settle a specifically defined dissension among four members of an
academic group. If, in each case, my friend gives voice to the thoughts
that successively come to mind, the chances are that we shall apperceive
the components of a constructive process operating before our ears from
the beginning to the end of the experiment — influxions of ideas from
the £CwellJ> of mind (What are they? How fast do they come? How varied
are they? How definite? How appropriate to the given task?), inter-
spersed with evaluations of these influxions, the rejection of some and
the acceptance of others (How much consideration is given to each
idea? How exacting is the standard of assessment? How excellent are the
judgments in the opinion of experts? How much inhibition, hesitation,
censorship, self-criticism occurs along the route? How quick are the
acceptances? How decisive?), and then, to make a long story shorter, the
temporal allocations, or ordinations, of the accepted components of the
design, the parable, or the plan (Are the concatenations actually logical?
Clearly expressed? Have all probable contingencies been met? Has any-
thing essential been omitted? How superficial or profound is the offered
solution or composition, and so forth). In every such experiment I sub-
mit, we shall obtain a unique mental composition which, at one extreme
and perhaps in the majority of cases, may be socially worthless in the
estimation of qualified judges and advisedly forgotten, but at the other,
might be a rare gem of creativity, something memorable that may
eventually find a place in the great body of cultural transmissions. We
may, for instance, be dealing with a Whitehead equal to such utterances
as these:

Insistence on clarity at all costs is based on sheer superstition as to the
mode in which human intelligence functions.

No science can be more secure than the unconscious metaphysics which
it tacitly presupposes.

Murder is a prerequisite for the absorption of biology into physics as
expressed in (its) traditional concepts.

Preparations for Scaffold of a Comprehensive System 43

A self-satisfied rationalism is in effect a form of anti-rationalism. It
means an arbitrary halt at a particular set of abstractions.

A science which hesitates to forget its founders is lost.

Scientists animated by the purpose of proving themselves purposeless
constitute an interesting subject for study.

Of course, creativity — the real thing — is an autonomous and capri-
cious process which rarely shows itself when called upon; hence, im-
promptu tests are not likely to bring forth anything but rather shallow
forms of originality and inventiveness. Nevertheless, to my way of think-
ing, there are compositional processes at work, ordering ideas and
shaping sentences — sometimes brilliantly — in the course of every com-
munication. Most of us, to be sure, make use of the same worn words
and trite phrases time and time again, and integrativeness in speech or
writing is limited to the joining of one commonplace to the next;
but were we to abide by the current laws of learning and in talks with
friend or spouse repeat tomorrow the response — the bit of news, the
joke, the idea — that was reinforced today, we would be heading for
press rejection or divorce. What we have to learn is to break a specific
speech-reward connection and on a subsequent occasion substitute some
variation. In short we wiU be rewarded only for saying something dif-
ferent from, but as stimulating as, that for which we were rewarded
last. Conclusion: a gust for novelty and emergent forms is widely
distributed among members of our breed.

For the present, we may define participant creative processes in
terms of their effect, result, achievement, namely, an unprecedented
form, and confine our attention to stable forms which are retrospectively
apperceived as valuable and as having further consequences in an
evolutionary context. Striking to many of us is the blindness of these
processes, their experimental character, and their resistance to the
coercions of conscious purpose, which is something that is worth con-
sidering in connection with human imaginations, and the occurrence in
some people of a strong disposition to create : to combine sounds, images,
words, concepts, propositions, ideas, ordinances, people, things, strategies,
or techniques in new and significant forms which express something that
is worth expressing, order things that are worth ordering, build some-
thing that is worth building, or solve a problem that is worth solving.
Mobilized by a need of any other class than this, a human subject is
likely to have a picture in his mind's eye of what he wants — water, sexual
intercourse, a habitation, an automobile, world news, membership in a
certain group, promotion, prestige, or what not. Under most circum-
stances, what he wants already exists somewhere, actually or potentially,
in the environment, and he must take it pretty much as it is or as it

44 HENRY A. MURRAY

comes. There Is food in that restaurant, information he requires in that
book, a person over there whose friendship he might win, a job to be had
and money to be earned, et cetera, et cetera. But the aim of creativity —
say, a design for a more efficient machine, an architectural innovation, a
symbolic plot for a drama to be written, the explanation of an enigmatic
phenomenon, a more enlightened foreign policy — has no existence any-
where. A person with this need must work, think, brood, daydream,
rest, sleep, turn his thoughts to other things — perhaps drink and read
detective stories — until his mind will favor him with a representation
which possesses, in his prospecting eye, the attributes that he seeks, and
then he must be favored further by representations of suitable embodi-
ments. A man may rack his brains throughout a lifetime without re-
ceiving the vision or idea for which he longs, or if the idea has come to
him, he may labor for years without finding the way to expound it in
a persuasive manner or to implement it in an actional endeavor. That is
to say, we are dealing here with energies of the human mind that do not
respond directly to voluntary efforts. Voluntary efforts can influence
their direction, defining, so far as possible, the target of their endeavor,
but they cannot force them to render up the desired form or answer.

Nowadays it is pretty generally agreed, I would suppose, that
imaginations of any real consequence are generated outside, or "below,"
the stream of awareness, after a more or less prolonged period of in-
cubation, and they are apt to leap to consciousness abruptly at the most
unexpected moments. Sometimes, like a dream, they seem to come from
without rather than from within the mind. A vision has been called a
vision because it is a visual presentation, a present, a gift, to the inner
eye, just as the heavenly constellations at night are a presentation, or
gift, to the outer eye. It was partly on this ground, we may surmise, that
the ancients believed that visions of import came from the gods, as best
among their blessings to deserving men. Today we are disposed to say
that they come from the unconscious. But the proposition I am sub-
mitting here is that the witting purpose to create something with certain
valued properties is almost wholly blind, its goal being to conceive a goal ;
and though voluntary effort is one determinant of success, the processes
on which creativity depends proceed, for the most part, spontaneously
and autonomously outside of consciousness and give rise to hundreds of
influxions which do not survive because consciousness rejects them, and
if a certain influxion is considered worthy of survival it may not be
what consciousness was seeking, but something else entirely.

Facts of this order constitute the basis for the not uncommon ex-
perience among creative men of serving as a vehicle or mouthpiece of
some supernatural or superpersonal imperative, of being an agent of
evolution instead of a feverish egoistic little self. "This is the true joy

Preparations for Scaffold of a Comprehensive System 45

in life," Bernard Shaw has written, "the being used for a purpose
recognized by yourself as a mighty one."

INFLUENCE OF SOCIAL EVOLUTIONISTS, CULTURAL
ANTHROPOLOGISTS, AND SOCIOLOGISTS

As one of the charter members of the Department of Social Relations
at Harvard, I could hardly fail to be inspired and directed in my think-
ing by our largely shared ambitious aim to advance by successive trials
toward a common theoretical system for basic social science. If it had not
been for this association, for the continuous influence of such learned
and persuasive colleagues as Clyde Kluckhohn and Talcott Parsons, I
might still be representing personalities in so near a social vacuum as we
did in Explorations. As a biologist I had been attached to the concept
of the herd instinct, as elaborated, say, by Trotter, and as a psycho-
analyst, to the concept of identification in each of its different meanings,
as well as to the several propositions respecting the internalization of the
parental superego. Nothing is more apparent as we look at others and
ourselves, especially in the United States — despite or because of our
loudly avowed ideology of freedom and individuality — than the tre-
mendous prevalence of unconscious imitation and conformity, of the
educing and constraining force of public opinion and behavior. But I
did not become aware of the numerous cultural differentiations one
had to make, differentiations of socioeconomic classes, of special sub-
groups, of rank in the decision-making hierarchy, of role and function,
until I gave a joint seminar with the encyclopedic Kluckhohn, who
consented to the office of tutor in these matters. There I once again
experienced the truth of the old adage : the best way to learn a subject
is to teach it, in this case in conjunction with an expert. Besides my
indebtedness to the elaborate classifications and generalizations of Tal-
cott Parsons, I should mention among other respected instructors in the
social sciences : Pareto as expounded by L. J. Henderson in a memorable
seminar, Malinowski, Sapir, Margaret Mead, Ralph Linton, John
Dollard, Florence Kluckhohn, Edward Shils, Robert Merton, Harold
Laswell, Ernest Cassirer — the list is long; many congenial influences
have necessarily been omitted.

Since the anthropological and sociological concepts that I employ are
pretty nearly all derivative, I need not say much on this score. Here
again I have been influenced by Darwin, specifically by the theory that
the group more than the individual has been the evolutionary unit. Being
of this persuasion, I have come to think that no theoretical system con-
structed on the psychological level will be adequate until it has been
embraced by and intermeshed with a cultural-sociological system. Al-

46 HENRY A. MURRAY

though every individual has some measure of inner life, a host of private
and largely secret feelings, fantasies, beliefs, and aspirations, and has
some extent of free play outside the coercions and restraints of the social
system, the great bulk of his overt behaviors are regulated by the mem-
berships and roles to which he is committed, his actual behavior being
the resultant of a fusion or compromise between cultural specifications
and standards and his own dispositions and abilities. Such is the con-
ventional doctrine of our time, in one guise or another, and I have little
to add to it. But, since the group theory of evolution is rarely mentioned
today and since, for better or for worse, it has strongly influenced my
speculations, I am yielding to the temptation of quoting a few para-
graphs from a recent attempt I made to expound it in a condensed
form.

Surveying the evidences of man's development on earth, the later
Darwin concluded: first, that the survival of the fittest is a principle
which applies decisively not so much to individuals as to rival groups —
tribes, states, or nations — and second, that mutual sympathy, aid, and
collaboration among members of a group are conducive to its solidarity,
and hence to its combative power and survival. To put it another way,
one of the critical variations established long ago was a clannish com-
bination of families more powerful than any single person, a flexible yet
stable social system with some differentiation of functions and conse-
quently with an enhanced capacity to cope with various tasks and
crises.

From the beginning, if we follow Sir Arthur Keith's composition of
the evidence, every successful group has adhered to a double code of
conduct, a Janus-faced morality: one face preaching submission to
authority, reverence, cooperation, loyalty, good will, and generosity
within the group, and the other more contorted face shouting with rage
and murderous aggression toward members of opposing groups. Other
things being equal, it must have been the clans or tribes which embodied
this dual standard in the best balance that triumphed and endured, and
passed on to their descendants down to the present day the dispositions
which sustained it.

This theory of group evolution helps us to understand why man is a
social, rather a solitary, self-sufficient creature and why, as a social
creature, he is both humane and brutal. Illustrative of his social prop-
erties are such familiar facts as these : that the vast majority of men are
reared in one particular society, a society that is prejudiced in its own
favor, and are satisfied to be lifelong interdependent members of this
society, that the bulk of their enjoyments come from interacting with
its members, that they are at peace with themselves only when they

Preparations for Scaffold of a Comprehensive System 47

feel and act in accord with its customs and ideals, and that, even in their
furthest reaches of self-forwarding ambition, they choose for their most
delectable final prize the applause of their fellow-beings, and after
death, fame, "that last infirmity of noble mind." The dual morality of
groups — tribes and nations — accounts, in some measure, for the failure,
the halfheartedness and insincerity, of all attempts to abolish war and
for the fact that human beings have been generally so willing, even eager,
to suppress their fears of self-extinction and fight for their country to
the tragic end, as well as for the fact that a man who kills a hundred
members of an enemy society is declared glorious, but is condemned to
the severest punishment if he stops the life of a single fellow citizen.

It is supposed that the generally victorious groups were those which
most fully incorporated and exploited the vaingloriousness and pride, the
greed and will to power of their individual members. But what is the
significance of the will to power? Power, intoxicating as it may be to
some men and to some nations, is a means to something, not an end.
Power for what? To this question the response of a creative evolutionist
might be: power to construct ever larger and less vulnerable social
systems controlling ever larger areas of the earth's resources, or in other
words, power, spurred on by greed, to grow and to develop, by invading,
conquering, subjugating, and assimilating weaker units, or more peace-
fully and happily in some cases, by federating with other units. History
reports a great number of such sequences: the integration of primal
groups into clans, and of clans into tribes, and of tribes into small nations,
and the integration of small nations into great nations that subsequently
broke apart, the rise and decline, the evolution and involution, of mighty
civilizations, as Toynbee has shown us, but as yet no orchestration of
state sovereignties into a world order, no political embodiment of that
dream of universal fellowship which centuries of idealistic men have
recommended to our hearts.

In short, everything I have said relative to formations, transforma-
tions, malformations, and deteriorations on the psychological level is
applicable in a general, though not specific, way to the level of group
dynamics.

COMPELLING NEED FOR COMPREHENSIVENESS

Although I was educated on the principle that limitation of aim is
the secret of success in science, and that the scientist is responsible for
particulars, it must be only too apparent to you that I have been tempted
to depart from the wisdom of this strategy by the dream of an all-
embracing scheme, a unified science, not, of course, to be achieved in my

48 HENRY A. MURRAY

own lifetime but in the distant future, if there is to be a future for our
species.

I suppose it would be proper to speak of :

1. A comprehensive concept (such as energy, process, matter, form,
motion) which refers to something that is always and everywhere ob-
servable or inferable.

2. A comprehensive conceptual scheme (such as the periodic table,
classifications of botanical and zoological forms) which differentiates
relationally all entities and all attributes and properties of entities within
the domain of a single discipline.

3. A comprehensive formulation, theory, or law (such as e = me2,
the laws of thermodynamics, the theory of evolution) which is applicable
over a wide range of phenomena.

4. A comprehensive spatial scope of individual concern within a
single discipline, such as (to limit consideration to the biological and
social sciences) that of a physiologist who takes the total organism as his
province (rather than specializing in kidney function), that of a psy-
chologist who takes the whole personality (rather than specializing in
cognition) , or that of a sociologist who takes the total community (rather
than specializing in family structure) . Scope of data.

5. A comprehensive temporal span of individual concern within a
single discipline, such as that of a biologist who is interested in genetics
and heredity, that of a psychologist who is occupied with parental as
well as subsequent determinants of personality, or that of a sociologist
or anthropologist who studies historic transformations. Span of data.

Now, one of the best appraisers of the status of psychological theory
in this country, the wisely chosen Director of this project, stated not so
long ago that the development of our science had been more retarded
in recent years by straining after comprehensiveness than by any other
variety of ambition. But since it is not clear to me which of the above
forms of comprehensiveness he had in mind, I have not yet had to
square my shoulders to the verdict guilty. There is at least one form
of comprehensiveness for which I have not reached, the comprehensive-
ness of a neat net of postulates and theorems that is expected to catch
every kind of fish that swims in the stream of human experience and
behavior. I have never been so optimistic as to think that we psychologists
were anywhere near the day when some master mind might achieve so
much. Instead I have been a perpetual catcher and collector of facts
and figures, a perpetual classifier of concepts, and a promoter, in a little
way, of marriages of concepts, believing that these pedestrian occupations
were appropriate to the stage of conceptual evolution at which psy-
chology has arrived. Here I am not speaking for the psychobiologists who
study the ways-means learning processes of imprisoned animals. They,

Preparations for Scaffold of a Comprehensive System 49

so far as I can see, have already arrived at that state of knowledge and
mastery of their variables from which law-making for their territory
makes sense.

The forms of comprehensiveness of which I am most surely guilty
are comprehensiveness of territory in space and time. I have spoken of
my interest in creative evolution down the ages and in developments of
personalities from birth on (temporal comprehensiveness) ; and I have
indicated how I was forced, to put it bluntly, by rny colleagues at
Harvard to become socio-spatially comprehensive, concerned with the
supraorganism of which every personality is imperatively a functioning
component. Nor can other groups, out-groups and foreign nations, be
excluded from the picture, it being all too evident these days that a little
shooting incident on some distant surface of our planet might initiate
a global conflict which would change the roles, the activities, and the
effects of millions of human beings. Belief in the imminence of a
catastrophic war is currently one of the determinants of anxiety in a
large number of people occupying statuses of responsibility. And then,
beyond the earth and its contentious nationalities, revolve the sun, the
moon, the planets, stars, and Milky Way, all of which have influenced
the minds of countless individuals and collectivities, not as the Chaldean
astrologers surmised, but by drawing aspirations and cognitions upward,
by engendering images and stories of celestial divinities and powers, of
resurrections and ascensions to a heavenly paradise beyond the grave,
and of life everlasting in a society of musical winged beings, not to
speak of the attraction by cosmic bodies of astronomers and poets.

TOLERANCE OF UNCERTAINTY

From what I have confessed so far it must seem as if the need for
certainty, powerful in most scientists, is very weak in me. But, as I weigh
them, my hopes and expectations in this regard are no higher and no
lower than they legitimately can be nowadays in the sphere of endeavor
to which I am committed. Were my demands greater, I either would be
perpetually defeated or, to escape from this, would be impelled to quit
personology and return to chemistry for peace. I take heart from
Aristotle: "It is the mark of an educated man to look for precision in
each class of things. . . .M

But this is not the whole story. There is something more in me which
is not irrelevant to this issue: the induction from experience that a
compulsive need for intellectual certainty — abetted, I would suppose, by
longings for personal security — is very apt to lead to deadly falsifications
and distortions of reality. Leaving aside the changeless eternal forms and
absolutes of philosophers and theologians, and confining ourselves to

50 HENRY A. MURRAY

scientists, we can find innumerable examples of the operations of this
need: the selection of the most fixed, permanent, or recurrent things
to study, the unnatural stabilization of the experimental environment,
the prevention of all but two or three possibilities of response, the
circumscription of the area of observation to a small part of the total
field of influential forces, et cetera, et cetera. Such choices and con-
straints are valid parts of the strategy of science and not criticizable
as such. They are to be criticized only when the results obtained in
highly focused studies of this sort are generalized across the board and
the notion propagated that the entities with which we are concerned
are far more structured, rigid, stable, orderly, consistent, and predictable
than they really are.

In my philosophy there are no absolute or inevitable laws, no en-
during certainties : every observation, every inference, every explanation,
and every prediction is a matter of less or greater probability. To this
most psychologists, I trust, would be ready to assent.

INTEREST IN SYSTEMS

My interest in systems was confined at first to shifts of equilibria,
as a function of oxygen tension, among the electrolytes of blood within
the walls of a glass vessel. The scope of the next system I studied with
some care was a volume bounded by an eggshell, closed to material sub-
stances but open to intakes and outputs of gases. Here my chief source
of illumination was Elements of Physical Biology by Lotka [2]. But the
relevance of these investigations and formulations to psychology was not
apparent to me until the thirties when I was introduced to Pareto's
representation of society as a system, and somewhat later to the con-
ceptualizations of the Chicago group as set forth, say, in Levels of Inte-
gration in Biological and Social Systems, edited by Robert Redfield [5].
Ever since, encouraged by Whitehead's speculations, I have been addicted
to the perilous practice of discovering analogies among events at dif-
ferent levels. This hobby, once private and covert, has become more
articulate of late, partly owing to parallels discovered in the writings of
L. von Bertanlanffy, A, E. Emerson, R. W. Gerard, and other men who
are concerned with correspondences and differences between various
kinds of systems — what is now known as General System Theory.

I am wary of the word "system," because social scientists use it very
frequently without specifying which of several possible different denota-
tions they have in mind; but more particularly because, today, "system"
is a highly cathected term, loaded with prestige; hence, we are all
strongly tempted to employ it even when we have nothing definite in
mind and its only service is to indicate that we subscribe to the general

Preparations for Scaffold of a Comprehensive System 51

premise respecting the interdependence of things — basic to organismic
theory, holism, field theory, interactionism, transactionism, etc. For ex-
ample, the terms "personality-as-a-whole" and "personality system" have
been very popular in recent years; but no writer, so far as I know, has
explicitly defined the components of a "whole personality" or of a
"system of personality." When definitions of the units of a system are
lacking, the term stands for no more than an article of faith, and is mis-
leading to boot, in so far as it suggests a condition of affairs that may
not actually exist. It suggests not only that one is dealing with a set of
recurrent, orderly, lawful interactions, but that the number, constitution,
position, and effects of the interacting units remain relatively constant.
That is, it is usually taken for granted that "system" refers to a homeo-
static, boundary-maintaining system. Finally, overtones convey the im-
pression that the speaker has a steady, coherent theoretical system in his
head which conforms to the steady coherent system he is studying.
Hence I am wary of the word. But, having found that I cannot get
along without it, I must do my best, when the time comes, to define
my restricted usages of this term.

I might say, in a general way, that, for me, system applies to a more
or less uniform integration of reciprocating and/or cooperating func-
tional activities, each of which, under favorable conditions, contributes
to the continuation of the entire cycle of activities which constitute the
system. As a rule, such a system is boundary-maintaining. According to
this view, each entity (form of matter) involved in a cooperating system
may be called an organ, relative to that system, each organ being defined
in terms of process and its contributing effect, or since organ processes are
not always capable of achieving a contributing effect, in terms of their
direction, endeavor, or intended effect. Thus, each unified, boundary-
maintaining system may be partially defined by representing the integra-
tion of successive processes and effects which are required to keep it
growing and/or to keep it going as a unique and vital whole. The major
unitary functional systems with which a social scientist is concerned
are these: personality systems, dyadic social systems, polyadic social
systems, representational (symbolic) systems, each of which may be
divided — according to different spheres of concern — into large sub-
systems. For example, a personality system may be divided into :

1. A psychosomatic system, consisting of all needs and activities
concerned with the growth and welfare of the body: procurement and
incorporation of water and food, transposition and allocation of food
particles, differential construction of frame and organs, excorporation
of water and waste, actuation and integration of muscular patterns,
development of manual and athletic skills, defense of the integrity of the
body, etc.

52 HENRY A. MURRAY

2. A psycho-material system, consisting of all needs and activities
concerned with the acquisition, restoration, or construction of a territory
and/or of a habitation (stead and shell), as well as with the acquisition,
restoration, or construction of implements or machines, utilization of
these implements, development of technical skills, defense of property,
etc.

3. A psychosexual system, consisting of all needs and activities con-
cerned with erotic love : stimulations and interactions, the formation and
continuation of an erotic dyad, conjugations, and the conception of off-
spring, etc.

4. A psychosocial system, consisting of all needs and activities con-
cerned with nonerotic social reciprocations: transmissions and receptions
of affection, of food, money, and material entities, of information and
evaluations, of orientations and ordinations, directions and compliances,
development of social skills, etc.

5. A psycho-representational system, consisting of all mental (cogni-
tive and ordinative) needs and activities associated with the above-
mentioned systems — acquisition of knowledge, explanations, and postula-
tions — as well as mental needs and activities concerned with impersonal
symbolic systems (explicit culture), with law, art, science, morals,
ideology, and religion, development of mental skills.

The personality system, as such, is concerned with the allocation of
time and energy among these different subsystems and sub-subsystems,
the ordination of their component serial endeavors, the repression of un-
acceptable emotions and impulses, and the reduction of conflicts and
strain.

A dyadic system consists of the interplay of two personality systems,
each of which is given equivalent attention. This is enough to indicate,
very roughly, the way the term "system" is applied in the scaffold as
now constituted.

APOLOGIA

When, after finishing part 1 of this assignment — my autobiography
of somewhat relevant cerebrations — I got round to a closer examination
of the scheme provided us, I discovered that it was even more exacting
than I had initially believed. It was definitely beyond my reach, beyond
the reach, I judged, of anyone who is primarily concerned, at this stage
of things, with the formulation of different types of personalities as
manifested, say, by different classes of reactions to a variety of similar
situations, rather than with the reactions of most people, say, to modi-
fications of one particular experimental situation.

I might have profited by the moral of the Icarian thema, as repre-

Preparations for Scaffold of a Comprehensive System 53

sented in the careers of several young persons assessed at the Baleen, an-
nex of the Harvard Psychological Clinic. Its moral is that of the inevitable
fall of over-reaching aspiration, the nemesis of hubris, so familiar to the
Greeks. But the prospect of this outcome did not bring about a reason-
able abandonment of the project. It merely served to check me to the
point of regarding the committee's standard as an unrealizable ideal,
but yet something to be held in view while I labored over the develop-
ment of the scaffold. As it turned out, the effect of this ideal was an
almost continuous procession of very general as well as of very particular
conceptual compositions, decompositions, and recompositions, which
kept informing me of the intricate influence of more and more variables
in the determination of the course and outcome of almost every unit of
behavior that could interest a personologist. Thus, I was led on from
complication to complication, and though many were resolvable, the
resolutions served only to increase the number of aspects to be con-
sidered and of discriminations to be made in analyzing, explaining, or
predicting any sequence of significant transactions. After a year or more
of this sort of thing, the produce of variables had reached an unmanage-
able degree of refinement and of magnitude; and, approaching the dead-
line, I was reminded of the judgment of Hippocrates: life is short, the
art long, occasion instant, decision difficult.

REFERENCES

It is not possible to pinpoint in the vast libraries of books and periodicals
the precise source of each assumption, concept, method that has been
mentioned in this paper. I have included the names — all well-known — of
the more influential theorists, but what I have acquired from some of
these — Henderson, Jung, Prince, Alexander, Whitehead, Lewin, Kluckhohn,
Allport, Parsons, and others — came very largely through conversations and
discussions, and what I have acquired from the writings of these and others
came, not from one article or book, but from pretty nearly all their works.
This is not the place, it seems to me, to list the relevant works of Aristotle,
Darwin, William James, Bergson, Lloyd Morgan, Santayana, Whitehead, or
Gassirer, or of such social scientists as Pareto, Parsons, Lasswell, Malinowski,
Sapir, or Kluckhohn, or of such authors as Janet, Freud, Jung, Adler, Rank,
Alexander, Horney, Sullivan, Kris, or Erikson, or of those psychologists who
are concerned with personality, such as McDougall, Allport, Murphy,
Maslow, Adams, or McClelland.

My constant disposition has been to select new fields of investigation and
to avoid those which have already been occupied, if not packed, by com-
petent experimentalists. For example, I have had no first-hand experience in
dealing with the intricate problems of perception or of animal learning, and
hence I have mentioned but few names of psychologists who have con-
tributed to our understanding of these phenomena. Many of these have in-

54 HENRY A. MURRAY

fluenced me directly as well as indirectly. But it would hardly be appropriate
in this place to list the works of such men as Pavlov, Thorndike, Watson,
Hull, Tolman, Bollard, Mowrer, Neal Miller, or Skinner, or of those who
have been concerned with the philosophy, logic, or semantics of theory
building, such authors as Bridgman, Stevens, Hull, Lewin, Koch, Egon
Brunswik, Else Frankel-Brunswik, Bergmann, Meehl, or Feigl.

So far as my own bibliography is concerned, the latest edition of it can
be found in C. S. Hall and G. Lindzey, Theories of Personality (1957),
published since the completion of all that I have written here.

The works referred to in the text are these:

1. Henderson, L. J. The order of nature. Cambridge, Mass.: Harvard
Univer. Press, 1917.

2. Lotka, A. J. Elements of physical biology. Baltimore, Md.: Williams &
Wilkins, 1925.

3. Murray, H. A., et al. Explorations in personality. New York: Oxford
Univer. Press, 1938.

4. Murray, H. A., & Morgan, Christiana D. A clinical study of senti-
ments. Published separately and in Genet. Psychol. Monogr., 1945, No.
32.

5. Redfieldj R. (Ed.) Levels of integration in biological and social
systems. Lancaster, Pa.: Jaques Cattell, 1942.

THE STRUCTURE OF PSYCHOANALYTIC
THEORY: A SYSTEMATIZING ATTEMPT1

DAVID RAPAPORT

Austen Riggs Center

Formerly I found it extraordinarily difficult to accustom my readers to
the distinction between the manifest dream-content and the latent dream-
thoughts. Over and over again arguments and objections were adduced from
the uninterpreted dream as it was retained in the memory, and the necessity
of interpreting the dream was ignored. But now, when the analysts have
at least become reconciled to substituting for the manifest dream its meaning
as found by interpretation, many of them are guilty of another mistake, to
which they adhere just as stubbornly. They look for the essence of the
dream in this latent content, and thereby overlook the distinction between
latent dream-thoughts and the dream-work. The dream is fundamentally
nothing more than a special form of our thinking, which is made possible
by the conditions of the sleeping state. It is the dream-work which produces
this form, and it alone is the essence of dreaming — the only explanation of
its singularity [S. Freud: The Interpretation of Dreams, 98, pp. 466-^-67].

Introduction 57

I. Background Factors and Orienting Attitudes {!}*. ....... 59

A. Background Factors 59

B. Orienting Attitudes 63

1 . The nature and limits of psychological prediction 63

2. Level of analysis 66

3. Utility and role of models 67

a. The reflex-arc (or topographic) model 67

b. The entropy ( or economic) model 68

c. The Darwinian (or genetic) model 68

d. The Jacksonian (or neural integration hierarchy) model .... 70

e. A combined model 71

'The completion of this study was aided by the Ford Foundation's grant in
support of research at the Riggs Center.

* The bracketed numbers, when they occur in the tables of contents of the essays in
this volume, indicate items in the Suggested Discussion Topics relevant to the headings
which they follow. See Note on the Use of Discussion Topic Index Numbers in the Appendix.

55

56 DAVID RAPAPORT

4. The comprehensiveness of empirical reference 78

5. Quantification and mensuration 79

6. Formal organization 82

II. The Structure of the System {2} 82

A. The Subject Matter of Psychoanalysis Is Behavior (the Empirical Point
ofView) 82

B. Behavior Is Integrated and Indivisible: the Concepts Constructed for Its
Explanation Pertain to Different Components of Behavior and Not to
Different Behaviors (the Gestalt Point of View) 83

C. No Behavior Stands in Isolation: All Behavior Is That of the Integral and
Indivisible Personality (the Organismic Point of View) 85

D. All Behavior Is Part of a Genetic Series, and through Its Antecedents, Part
of the Temporal Sequences Which Brought About the Present Form of the
Personality (the Genetic Point of View) 86

E. The Crucial Determinants of Behaviors Are Unconscious (the Topo-
graphic Point of View) 88

F. All Behavior Is Ultimately Drive Determined (the Dynamic Point of
View) 89

G. All Behavior Disposes of and Is Regulated by Psychological Energy (the
Economic Point of View) 91

H. All Behavior Has Structural Determiners (the Structural Point of View) 93

I. All Behavior Is Determined by Reality (the Adaptive Point of View) . 97

J, All Behavior Is Socially Determined (the Psychosocial Point of View) . 101

K. Discussion 104

III. The Initial Evidential Grounds for the Assumptions of the System and
Their Strategic Character {3} 110

A. Initial Evidential Grounds 110

1 . The assumption of psychological determinism Ill

2. The assumption of unconscious psychological processes 112

3. The assumption of unconscious psychological forces and conflicts . . 112

4. The assumption of psychological energies and their drive origin . . 113

B. Strategic Choice of Initial Evidential Grounds 114

C. The Relation of the Observations to the Theory 116

IV. Construction of Function Forms {4} 121

V. The Problem of Quantification {5} 124

A. Cathexes 125

B. Dimensional Quantification 129

VI. The Formal Organization of the System {6} 133

A. The Present Status of the System 133

B. The Desirable Level of Formalization 135

VII. The Range of the System's Applications {7} 136

VIII. History of the System's Research Mediation {8} 138

IX. The Evidence for the System {9> 140

The Structure of Psychoanalytic Theory 57

A. Current Status of Positive Evidence 140

B. Major Sources of Incompatible Data 143

C. "Critical" Tests of Principal Assumptions 148

X. Methods, Concepts, and Principles of Broad Application {10} . . . . 149

A. The Range of Application 149

B. Methods, Concepts, and Principles of Long-term Significance . . . 150

1. Methods 151

2. Principles 152

3. Concepts 153

a. Dynamic point of view 153

b. Economic point of view 153

c. Structural point of view 1 53

d. Genetic point of view 154

<?. Adaptive point of view .... 1 54

XI . The Theory's Achievements and Its Convergence with Other Theories {1 1} 155

A. Achievements 155

B. Convergence with Other Theories 157

XII. Tasks for the Future Development of the Theory {12} 159

A. Empirical Evidence Needed 159

B. Obstacles to the Development of the Theory 161

C. The Practical Obstacles to Theoretical Advance in Psychology . . 163
References 167

INTRODUCTION

Neither Freud's nor other psychoanalysts' writings give a systematic
statement of the psychoanalytic theory. This fact, combined with my
acceptance of the outline suggested by Dr. Koch (reflected in my
section headings), imposed problems that the writers of the other essays
may not have had to face. It is proper, therefore, to state the premises
of this essay.

1. Freud's writings are the source of psychoanalysis and provide the
frame of reference for its systematic treatment. Thus this essay centers
on Freud's work.

2. A systematic treatment of the theory should also take into account
other contributions which decisively shaped the present form of the
theory. Thus this essay draws extensively on Hartmann's and Erikson's
work.

3. The systematic statement of the theory should establish its rela-
tion to the alternative ("Neo-Freudian") theories which arose from it.
But an early attempt at systematization, such as the present one, can
neglect them without prejudice. Thus this essay barely touches on Adler,
Jung, Rank, Homey, Kardiner, and Sullivan.

58 DAVID RAPAPORT

4. An attempt at systematization should stay as close as possible
to the formulations of the literature, but it should also interpret these.
This essay, therefore, although it hews close to the existing theory, does
draw inferences and does make judgments. Consequently the theory
it presents may appear unfamiliar to the reader whose conversance with
psychoanalysis is exclusively clinical or only cursory.

5. A systematic statement need not follow the emphases of the
literature. Its emphasis should be dictated by systematic considerations.
Thus this essay only touches on the theory of symptoms, psychosexual
development, therapy (e.g., transference and resistance), and con-
centrates heavily on what Freud called metapsychology. It makes a
distinction between what might be called the special or clinical theory
and the general or psychological theory of psychoanalysis.

6. One of the aims of Dr. Koch's outline was to make the essays of
these volumes comparable. To fulfill this requirement I found it necessary
to present some considerations (e.g., on independent, intervening, and
dependent variables, as well as on quantification) which have no direct
roots in the psychoanalytic literature and which enter frames of ref-
erence somewhat alien to my own thinking.

7. Dr. Koch suggested that the authors of these essays assume the
reader's familiarity with previous statements of the theory and dwell
primarily on systematic issues. Complying with his outline made some-
thing of this sort unavoidable. Yet I had to conclude, from recent writ-
ings of psychologists about psychoanalysis, that familiarity with the psy-
chological theory of psychoanalysis (as distinguished from the psycho-
analytic theory of neurosis) cannot be generally assumed. The historical
relationships — which play an important role in all unsystematized the-
ories— seem to be particularly unfamiliar. Thus, time and again, I found
it necessary to summarize theories and to sketch historical relationships.
The result of my attempt to reconcile these conflicting demands is not a
happy one. In the beginning of the essay the reader will find familiarity
with many concepts and theories taken for granted, only to encounter
some of them later on, again and again discussed in detail, with further
information added each time. The time limitation — unavoidable in such
collective endeavors as these volumes — permitted me no better solution ;
it is also responsible for the length of the essay. Had I prepared it on my
own schedule, it would have matured for a few more years, and it might
have become more comprehensive and tighter in its structure and "log-
ical joints.53

To my mind it is too early to attempt a systematization of the
psychoanalytic theory. A science can be a "good science" without being
ready for a systematic presentation: all old sciences were once in this
position. The existence of this essay is thus in need of explanation. I was

The Structure of Psychoanalytic Theory 59

prompted to write it partly by the urgings of Drs. Gill, Hartmann, Holt,
Klein, and last but not least, Dr. Koch, the coordinator of this APA
project, and partly by my wish to pave the way for an adequate
systematic presentation of psychoanalysis.

The very prematurity of this attempt had curious consequences. The
essay presents several cross sections (for instance, models, points of
view) of the theory which, though they are linked by identical concepts
and by common empirical referents, are not systematically related to
each other. The clearest indication of prematurity is the uncertainty
whether we are not yet able to connect these systematically, or whether
they need not or cannot be connected.

Since the literature directly bearing on the system of psychoanalytic
theory is meager, I refer — contrary to custom — to mimeographed ma-
terial of limited circulation and even to unpublished manuscripts. The
English Standard Edition of Freud's writings is still incomplete, therefore
the references are to that medley of editions which I have used over the
years in my studies. Some of these involve inaccuracies corrected by the
Standard Edition. While I am aware of these, I did not attempt to cor-
rect them.

The contradictions between this survey and the Rapaport-Gill study,
which went to press since this manuscript was prepared, are explained
partly by the survey character of this study and partly by the time lag.

As much as space permitted, I have referred to sources and acknowl-
edged the specific help I received. Drs. M. M. Gill, R. R. Holt, G. S.
Klein, and R. Schafer read the manuscript, and their suggestions and
corrections were so numerous that without a heavy addition of footnotes,
this is the only way I can acknowledge my indebtedness to them. To Dr.
Holt I am particularly grateful, not only for his repeated readings,
suggestions, and criticisms, but also for the share he had in shaping
the considerations on variables and quantification. But I am in even
greater debt to Erik Erikson, Merton Gill, Heinz Hartmann, and Samu
Rapaport. Last but not least, I want to express my gratitude to Mrs.
Ruth Shippey, Mrs. Barbara Kiley, and Miss Suzette Annin. Mrs.
Shippey and Mrs. Kiley did the secretarial work on the several versions
of this manuscript, and Miss Annin did the editorial and bibliographic
work.

I. BACKGROUND FACTORS AND ORIENTING ATTITUDES

A. Background Factors

The formative influences in Freud's background were the Jewish
tradition, an early developed interest in literature (particularly a de-
votion to Goethe and, through him, to ancient Rome), courses with

60 DAVID RAPAPORT

Brentano of act-psychology fame, the impact of Darwin's theory of
evolution, clinical and laboratory research in neurology and neuro-
anatomy (in the orbit of men from Helmholtz's circle), clinical psy-
chiatric work (with Meynert), clinical work with neuroses (at first with
Breuer, Charcot, and Bernheim), and self-observation.2

The influence of Helmholtz on Freud's theory is seen in the postulate
of thoroughgoing determinism, in the central position of the pleasure-
pain principle (and the primary process) which is patterned on the
concept of entropy,3 in the reality principle (and the secondary process)
which is patterned on the principle of least action, and in the "economic
principle" which is patterned on the principle of conservation.

The experience in neurological research is responsible for Freud's
conception (derived from Hughlings Jackson's view of the nervous
system) of a series of psychological organizations (instances, structures)
hierarchically and topographically superimposed upon each other. That
experience is also responsible for the conception of associative networks
organized superficially by contiguity but fundamentally by drives, for the
conceptions of inhibition and facilitation, at first bodily transported into
his system from neurology, and for his early assumption that psycho-
dynamics is neurodynamics. Even when abandoned, this assumption still
lingered on in the form of the belief that sooner or later psychodynamics
would be placed on the "solid footing" of neuro- and/or biochemical-
dynamics.

But Freud's laboratory research was also closely related to the theory
of evolution, and it is probably this conjunction which is reflected in the
genetic cast of Freud's thinking, particularly in the close relation hypo-
thesized between phylogenesis and ontogenesis,4 in the emphasis on
epigenesis, in the regression concept and many others. A Neo-Lamarckian
version of evolution theory also seems to have influenced Freud's
thinking [115, p. 64].

The effects of his clinical psychiatric experience with Meynert and
his related readings (e.g., Greisinger), though probably crucial, have
not been studied in detail.5 It seems reasonably certain, however, that
the contents of the hallucinations in "Meynert's amentia" served as the
prototype for the concept of "wish-fulfillment" [cf. 35, p. 136; 98, pp.

2 This list represents a narrow view of "formative influences." For a broader,
more psychological one, see Erikson [63, 64, 65] and Gross [154]; see also [15, 16,
17, 18; 193; 309].

3 Freud refers it to Fechner [69, sec. 11, p. 94, note]; see [123, pp. 3, 4].

4 Dr. F. Schmidl (Seattle) calls attention to Haeckel's particular influence.

5 But see Hartmann's recent study [165].

The Structure of Psychoanalytic Theory 61

509, 533] and as the foundation for what will be described below
as the primary model of cognition. It is also probable that the often
claimed influence of Herbart [see 193] and the less noticed one of
Hering6 on Freud came indirectly through the work with Meynert and
related reading. Neither his neurological research (dissection and micro-
scopy) nor his clinical psychiatric work provided Freud with experience
in the experimental method: both fostered his bent toward observation.

The experience with neurotics left its trace on Freud's system in the
recognition that psychopathological phenomena, such as 'the unana-
tomical delineation of hysterical symptoms, are organized on principles
other than those familiar to neuroanatomy and neurophysiology; in the
recognition of the power of psychological forces (through the observation
of hypnotic and waking suggestion with Breuer and Bernheim) ; in the
recognition of the existence of nonconscious psychic formations (through
observations of hypnosis and alternating states of consciousness by
Charcot and Janet), and the conceptualization of these as the System
Unconscious; and in the recognition of the crucial role of sexuality in
neuroses (Freud attributed his first inkling of this to Charcot's, Chrobak's,
and Breuer's incidental comments) .

The influence of self-observation (including his self-analysis) is
ubiquitous in Freud's theory, and accounts for the method of free as-
sociation, for the role of dream interpretation as an investigatory tool,
and for many specific discoveries.

The traces of Brentano's act psychology7 are less obvious and have
never been explicitly discussed.8 Yet the central position of instinctual
drives in Freud's theory parallels Brentano's interpretation (which con-
trasts sharply with that of Anglo-Saxon empiricists) of both stimulation
and response in terms of acts of intending. In the early phases of Freud's
ego psychology, Brentano's influence seems even more striking. The term
intention crops up, the problem of reality testing leads to an analysis of
the "belief in reality" [119, p. 146] along Brentano-like lines, and the
distinctions between what is perceived and what is conceived, what is
real and what is only thought, etc., come into play. This influence

fl Ernst Kris (personal communication, Jan. 11, 1957): "I have noted one of
the most obvious sources for Freud's thinking, namely Hering's paper on memory.
The evidence of Freud's interest reaching up to 1922 is absolutely conclusive and
as far as I know never noticed. It might amuse you to look in this connection
at Anna Freud's translation of Levine's book on the Unconscious. The translation
of the chapter on Butler is by Freud and so is an interesting footnote."

7 Concerning Freud's contacts with Brentano, see Merlan [232, 233].

8 Dr. F. Schmidl suggests that it was through Brentano that Freud came to
know of Maudsley, to whom he refers in The Interpretation of Dreams, and whose
concept of the unconscious may have influenced Freud's.

62 DAVID RAPAPORT

pervades the Papers on Metapsychology [108, 110, 114, 115, 116, 117,
119, 120]. And although Freud deliberately refused to have anything to
do with philosophy, he did acquire some familiarity with it through
Brentano. In one of the few specific references he makes to philosophy,
he characterized psychoanalysis (and particularly its concept of un-
conscious determination) as the psychological counterpart of Kant's
philosophical views [117, p. 104]. Indeed, the epistemological implica-
tions of psychoanalysis are closest to Kant [see 262] and most remote
from Anglo-Saxon empiricism.

The influence of literature in general and Goethe in particular are
again hard to trace. They certainly shaped Freud's interest in and grasp
of human nature. They provided the pattern for the case history as a
tool, which medical case histories of his time did not supply (compare
the best of these, Charcot's, for example, to Freud's). Indeed, it might
be said that the intrinsic validity9 of his reasoning and descriptive writing
often had to serve him as that indicator of validity which in older sciences
is usually provided by quantitative measures. He became one of the out-
standing masters of thought and pen in the German language (Goethe
Prize). These influences also fostered in him that sensitivity to the
subtleties of verbal communication and that readiness to seek meaning
behind meaning which, combined with a knack for metaphor and
symbol, are the requisites of interpretation. Indeed, they probably guided
him to his central conceptions — motivations, affects, and conflicts —
which are the raw material of all art.

The role of the Jewish tradition in Freud's thinking, methods, and
theorizing has not been explored in detail either. Wittels [327], Reik
[286], and Erikson [63, 64, 65] have elucidated some aspects of it. It
is possible that much of what we attribute to Freud's interest in literature
comes from the tradition of "the people of The Book." Associative and
interpretive methods have some of their most striking archetypes in the
methods of the Talmud. The stereotyped Aramaic phrase, introducing
Talmudic interpretation, translated into English reads: "What does he
want to let me hear?" But the degree of Freud's direct conversance with
his Jewish tradition and its effect on his thinking have not yet been
documented.91

^9By intrinsic validity I mean what literary criticism means when it speaks of
a "valid statement" : the great writer achieves a form which makes the expression
of his observations, feelings, and thoughts a "valid statement." But even in every-
day life, some of us convey an experience so that it is clear, convincing, and
pregnant with meaning, while the reports of others are pale, pointless, and diffuse,
as if they were third-hand.

9a In the period of reading proof of this article, I noted advertisements for an
apparently pertinent publication (D. Bakan. Sigmund Freud and the Jewish
Mystical Tradition. Princeton: Van Nostrand).

The Structure of Psychoanalytic Theory 63

We cannot abandon this survey of background factors without point-
ing to one whose significance has so far not been explored, the Zeitgeist
[see 31, 32]. Darwin and Helmholtz were certainly part of it. Freud's
Jewish background and fate, and their influence on his thinking, have
never been discussed in the broad terms of Zeitgeist. Only Erikson10 has
discussed the influence of the general atmosphere of Victorian Vienna,
which was probably as much a limiter of Freud's social psychology as a
source of his emphasis on sexuality and its vicissitudes. The assessment
of the Zeitgeist attains particular urgency because of the fundamental
similarities of Freud's theory to one immediately preceding it, and one
immediately following it in time. Marx, Freud, and Einstein, who con-
tinued the Copernican, Kantian, and Darwinian revolutions, relativized
our conceptions of the world. Marx, reversing Hegel's dictum, asserted
that "man's [economic] existence determines his consciousness and not
his consciousness his existence," and thus made man's view of his world
relative to his socioeconomic status. More broadly, Freud asserted that
man's view of and relation to his world are dependent upon (relative
to) his impulses and are not simply imprinted on him by his experience.
Most broadly, Einstein asserted that observation is relative to the ob-
server's position. If it should turn out that the commonality of the three
theories is as real as it seems, and is rooted in the Zeitgeist, then we
would have before us a background factor which, though subtle and
nonspecific, might prove the most pervasive and most powerful of all.

B. Orienting Attitudes

1. The nature and limits of psychological prediction. Prediction in
psychology implies the postulation of thoroughgoing determinism in
human behavior. Freud's assumption of exceptionless psychological
determinism, which is perhaps too easily taken for granted today, pro-
vides the necessary foundation for prediction.

Since the empirical material first dealt with was the already present
neurotic symptoms, Freud's primary causal problem was postdiction
rather than prediction. This initial situation is not unique to psycho-
analysis. It has its counterparts in the social sciences, e.g., in history,
and in the natural sciences, e.g., in the theory of evolution. A theory is
not invalidated by being postdictive, as long as postdiction is carefully
distinguished from ex post facto explanation.

Because the observations were made in the therapeutic situation, the
predictions were of necessity related to the effects of therapeutic inter-
ventions and thus were fraught with the same difficulties which have

"Erikson [64] also calls attention to the influence that the economic theories
of the time seem to have had on Freud's thinking.

64 DAVID RAPAPORT

beset recent investigations involving "participant observation" and
"action research." Freud believed that only first hand experience with
the psychoanalytic method of treatment could make understanding and
valid tests of psychoanalytic predictions possible, that the psychoanalytic
theory could be validated only by the psychoanalytic method, and that it
had, indeed, been so validated and needed no further validation. The
method, the theory, and its validation were considered inextricably tied
to one another.11

While these limitations on verification and prediction were pro-
fessed, from the beginning observations of hypnotic phenomena were in-
voked to validate the propositions concerning the unconscious [19, 35].
Hypnotically induced dreams [306], hypnagogic [310] and daydream
phenomena [319] were hailed as independent evidence corroborating
the predictive (and postdictive) power of the theory. Moreover, eth-
nology [109], literature [102], and psychotic products [107] were in-
creasingly invoked as independent confirming evidence, though their use
to corroborate the theory tended to merge with the use of the theory for
their explanation.12

More recently, it became increasingly evident that direct studies of
infant and small-child behavior were needed for the independent con-
firmation of the theory's postdictive reconstructions of these phases of
life, and many such investigations were undertaken. Psychodiagnostic
and experimental evidence has also been increasingly invoked as con-
firmation of the theory, though the investigations by which this evidence
has been obtained have rarely shown due regard for the complexity of

11 The discussion of "critical tests," in Section IX.C. below, points out that
whereas in other sciences tests validating a theory decide between alternative and
mutually exclusive possibilities, as a rule this is not possible for psychoanalytic
theory. The alternatives envisaged by psychoanalysis are not mutually exclusive
but rather equivalents which can substitute for each other, according to the dy-
namics of the situation. Thus the theory is not built by tests of predictions ex-
cluding all but one of several alternatives, but rather by the inclusion of all
observed alternatives which are consistent with the existing theory. Only those
alternatives which clash with the existing theory are excluded. The observation
which suggests such incompatible alternatives is rechecked by further clinical
observations. Rechecks which confirm the incompatible alternatives, and thus do
not permit their exclusion, lead to the modification of the theory. It is thus that
postdiction — guided by the aim of preserving the internal consistency of the theory,
rather than by the principle of parsimony — becomes the principal means of theory-
building in psychoanalysis.

12 All sciences must subject observations to interpretation in order to establish
their evidential significance for the theory. This is particularly conspicuous in
psychoanalysis, where the concepts are by and large at a considerable distance
from the observations. For a further discussion of this point, see pp. 116 fL

The Structure of Psychoanalytic Theory 65

the theory, and their methods have rarely been such that the results
could pass as confirmations of the theory.13

Psychoanalysis did concern itself with one class of predictions,
namely, prognoses. These are part of the subject matter of applied
psychoanalysis (clinical psychoanalysis and psychiatry) and not of
theoretical psychoanalysis proper. The problem of prognosis has three
facets: the prognosis for treatment by the psychoanalytic method, the
prognosis for "spontaneous remission," and the prognosis for treatment
by modified psychoanalysis or other therapy. So far the study of the
criteria of prognosis has yielded rules of thumb rather than theory,
yet the concepts of "ego strength" [158], "model technique" and "param-
eters of technique" [5 1] did arise in this context

Once the postdictive character of psychoanalytic propositions is clear,
another characteristic of the theory also becomes obvious. The detailed
study of dreams, of symbolism, of slips of the tongue, of wit, of as-
sociation sequences, and the like, suggests that psychoanalysis studies
and predicts behavior on this "microscopic" level; yet the actual aim
of the theory was always to predict or postdict life-sized ("macroscopic")
segments and sequences of behavior.14 This curious duality is char-
acteristic of the theory: it is holistic, but not because it lacks methods for
studying and predicting the "microscopic"; and it is atomistic in the
sense that it can and does study the "microanatomy" of behavior, but not
because its methods and interests limit it to "microscopic" phenomena.
Naturally, the verification of its theory of slips of the tongue by post-
hypnotic suggestions [53] or the verification of its theory of symbolism
by means of suggested dreams [306, 288, 244, 83], which involve "micro-
scopic" predictions, does not verify the "macroscopic" relationships pre-
dicted by the theory; in turn, verification of macroscopic relationships
(e.g., that of homosexuality to paranoia [cf. 246]) does not necessarily
confirm the detailed mechanisms (such as projection) which, according
to the theory, mediate these macroscopic relationships.

In conclusion : the nature of the material Freud worked on led him
to overemphasize postdiction and underemphasize prediction in building
his theory. In this he was also influenced both by the type of neurological
work he and his teachers pursued, and by the methods of the biological
science of the time. But it may be questioned whether or not any science
in its beginnings has been free from such imbalances. The basic necessary

13 The trouble with these investigations is that either their status as a source of
independent evidence for the theory is not established, or their relevance to the
theory is not established. Cf. Section V., below.

14 The terms "microscopic" and "macroscopic" are used here in the sense indi-
cated by the examples, without reference to any of their various usages in the
literature.

66 DAVID RAPAPORT

condition for predictions and for their confirmation is present in the theory
of psychoanalysis, and certain types of psychoanalytic predictions have
been confirmed. Moreover, postdiction, if properly handled,15 is as valid
a confirmation of a theory as prediction. The task ahead is to add to the
necessary conditions of prediction the sufficient conditions^ by tighten-
ing the theory and by developing adequate methods of quantification and
confirmation.

2. Level of analysis. The level of analysis has changed repeatedly in
the history of psychoanalytic theory.

First, Freud (1895) made an attempt [94] to account for all be-
havior by neurodynamics, though even in this period he already had a
clear outline [35] of his psychological theory, which centered on the
conflict between environment and ego (memory of traumatic experience
vs. social propriety and self-respect). At this point, he equated the ego
with consciousness (i.e., the dominant ideational complex) and the un-
conscious with what the environment disapproved of. Thus, early psycho-
analysis operated with three "levels of analysis": neuroanatomy and
neurodynamics, environment vs. ego, Conscious vs. Unconscious.

Second, in the next phase (1900) of the theory [98], "intrapsychic
dynamics," centering on the drive vs. censorship conflict, becomes the
causal referent of all behavior and the ultimate causal factor. But even
in this period censorship and secondary process are connected by Freud
with reality and interpersonal relations (environmental and psychosocial
referents). Yet the dominant level of analysis is the intrapsychic one, in
terms of drives vs. censorship.

Third, with the development of ego psychology (1923), a dual intra-
psychic reference system crystallizes [126]: drives and structures are
juxtaposed. The dominant level of analysis is still the intrapsychic one, in
terms of drives vs. structures.

Fourth, (1926) the structural concepts are recognized in part as
representing external reality referents17 and the drives are recognized as
representing biological referents.18 Thus the intrapsychic reference system

15 The difficulties in confirming postdictions are these: the data on which a
postdiction is based must in some inferable form imply the relationships to be
postdicted; however, the relationships implied in the data must not be so obvious
as to make postdiction superfluous. "Proper handling" of postdiction thus has to
make explicit both what is given in the data on which the postdiction is to be
based and what is not given and can be only inferred by postdiction. This is
easier said than done, however.

13 See Benjamin [1 1] for the first discussion of this issue in the literature.
See also footnote 20.

17 Cf. Section II. H., below.

ts Freud wrote: " . . . 'instinct' appears to us as a borderland concept between
the mental and the physical, being both the mental representative of the stimuli

The Structure of Psychoanalytic Theory 67

is reduced to organism vs. external reality, and a variety of hypothetical
constructs (drives and structures) are interpolated. There are now three
levels of analysis : biologic, intrapsychic, and real, though all of these are
handled in terms of their psychological representations.

Fifth, (1937-1946) the psychosocial referents crystallize in the
work of Horney [181, 182], Kardiner [194, 195], and Sullivan [313]
on the one hand, and in that of Erikson [56, 57, 59, 60] and Hartmann
[157] on the other. A system of multiple levels of analysis evolves, in-
cluding the dynamic, economic, structural, genetic, and adaptive levels,
whose foundations had already been built in the earlier phases.

In conclusion: the psychoanalytic theory, by its conception of "over-
determination," kept itself open to all relevant "levels of analysis," and
was not limited to a single one as were many other theories. Yet the
"intrapsychic" concepts in general, and the drives in particular, remain
central to the theory.

3. Utility and role of models. Freud's theory contains four distinct
models. They are united in the theory itself, but not in one single model.
We will first present each of these, and then attempt to develop a com-
bined model.

a. The reflex-arc (or topographic] model. This model [98, pp. 498ff.]
represents — as it does in the stimulus-response theories, too — the tend-
ency of the organism to respond to stimulation. The Freudian model,
however, has additional specifications :

1. This tendency is regarded as a direction of psychological processes.

2. It is one of the two directions excitations can take, the other being
the regressive.

3. In the ideal case the excitation begins in a sensory stimulation,
passes through the Systems Unconscious, Preconscious, and Conscious,
and terminates in motor action: this is the "topographic" course.

4. Not every excitation, however, need pass through the complete
topographic sequence.

For instance, excitations can originate in the Unconscious: drive-
excitations usually do so, though drive action is often triggered by a
stimulus. Excitations can also originate in the Preconscious : dreams are
initiated by preconscious day-residues. Nor is it necessary that excitations
initiated by a sensory stimulus run the whole topographic course; they
may terminate, temporarily at least, in the Unconscious or Preconscious :
that this is the case with "unconscious" and "preconscious" perceptions,
which are clinical commonplaces, has been confirmed by the experi-
ments of Poetzl [257], and others [e.g., Huston, Shakow, and Erickson.

emanating from within the organism and penetrating to the mind, and at the same
time a measure of the demand made upon the energy of the latter in consequence
of its connection with the former" [115, p. 64].

68 DAVID RAPAPORT

188; Diven, 47; Fisher, 84; Klein et al., 201]. Similarly, an excitation
can terminate in consciousness without initiating a motor response.
Recent developments in psychoanalytic ego psychology demonstrate that
autonomous functions of the ego (particularly automatized ones) may
short-circuit the topographic course. Thus a place within psychoanalytic
theory is allocated to automatized (one-to-one) stimulus-response rela-
tions. The topographic locus of origin is an important characteristic of
excitation processes.

This model was useful in two ways. On the one hand it coordinated
descriptively a welter of otherwise disparate observations, such as the
vicissitudes of stimulations, the alternative (ideational, affective, action,
and abeyant) responses to stimulations, the lack of one-to-one relation-
ships between stimuli and responses, and the wide variety of apparently
"spontaneous" ideational, affective, or action responses (ranging from
dreams, daydreams, delusions, blushing, sweating to parapraxes and
random movements) . On the other hand, it served as the foundation
for the topographic point of view in general, and for the concepts of the
Systems Unconscious, Preconscious, and Conscious in particular, and
these in turn were the predecessors of the structural point of view.

b. The entropy (or economic] model. This model [98, pp. 509, 533]
— implicit in the direction attributed to the course of excitation in the
topographic model — is the crucial, topographically incomplete19 sequence
of infant behavior: restlessness — > sucking on the breast -> subsidence
of restlessness. This sequence, which makes behavior the referent of
tension-reduction processes, is regarded as the basic model of all moti-
vated behavior, and — in keeping with the postulate of determinism —
pertains to obviously motivated behaviors as well as to apparently ac-
cidental ones. It can be modified — as we shall see — to account for
tension-maintaining and tension-increasing processes also. The merit of
this model is that it coordinates a wide range of phenomena, and serves
as the foundation for the concepts of the pleasure principle and wish-
fulfillment in particular, and the economic point of view subsuming them
in general. It plays an important role in the transformation of the topo-
graphic into the structural point of view, and also contains the core of
the dynamic and adaptive points of view. Since this model already im-
plies some of the others, we will later present a sketch of a previous
attempt [267] to develop it into a unified psychological model of psy-
choanalytic theory.

c. The Darwinian (or genetic] model. This model [101; cf. also 1],
which asserts that the course of ontogeny abides by inborn laws, served
Freud as the frame of reference for systematizing the data of his patients'

19 See p. 67, above.

The Structure of Psychoanalytic Theory 69

life histories, and became the foundation for the genetic point of view in
general, and for the theory of psychosexual (libido) development, in-
cluding the concepts of fixation and regression in particular. Freud's
inclination to alloy the Darwinian model with Haeckel's biogenetic law
(ontogeny repeats phytogeny) on the one hand, and with the Lamarckian
view of evolution on the other, led him to some inferences which have
been seriously questioned. Yet some of these inferences proved fertile, use-
ful, and independent of the grounds they were built on. Haeckel's bio-
genetic law helped Freud in elaborating and using the Jacksonian model.
Lamarck's model enabled him to conceive of processes of adaptation
for which Darwin's theory did not provide the conceptual means. The
core of Freud's genetic conception, namely psychosexual development, is
probably the most familiar segment of psychoanalytic theory, and we
need not dwell on the evidence which led Freud to make it the center
of his genetic model. The model's usefulness was not exhausted by Freud :
both Hartmann's [157] concept of "change of function" and Erikson's
[58, 62, 66] "epigenetic" conception (which extends the postulate of
"the lawfulness of ontogeny" to behavior development far beyond the
confines of psychosexual development) are based on it.

But this sketchy statement does small justice to the pervasive
significance of the genetic model in Freud's theory [see Rapaport, 279].
Actually, concepts as high in the theoretical hierarchy as identification
and transference, and theories as complex as that of object choice have
their roots in this model. It was the genetic model which enabled
psychoanalysis — unlike contemporary learning theories — to put, instead
of prior learning, prior inborn givens in the center of its conception of
learning. [For similar attempts, see Lorenz, 225; Tinbergen, 315; Piaget,
254; Schiller, 303.] Such genetic considerations made it possible for
Freud to realize the significance of early experiences for adult behavior.
It took academic psychology fifty more years to come to this realization.
[Cf. Hunt's confirming experiment, 186, 187, Hebb's theory, 169, and
Beach and Jaynes' review, 10].

Erikson's [61] as well as Hartmamrs [157] and his collaborators'
[167] work has advanced our genetic understanding, as has Hartmann
and Kris's discussion [166] of the genetic and the dynamic propositions
of psychoanalysis. Werner's [322] and Piaget's [254, 255, 256] work in
genetic psychology were advances in the same direction. Normative and
longitudinal studies have contributed considerable systematic observa-
tional material concerning genetic sequences. Yet the methodological
problems involved in the study of such sequences and in the application
of the genetic point of view have still not been solved.20

20 See John Benjamin, "Prediction and Psychopathological Theory," in press.

70 DAVID RAPAPORT

d. The Jacksonian (or neural integration hierarchy] model. Accord-
ing to this model, the nervous system consists of a hierarchy of integra-
tions in which the higher ones inhibit or control the lower, and damage
to or suppression of the higher ones reinstates the function of the lower.
When Freud abandoned his neurological anchorage (1898), he ceased
pursuing neuropsychological speculations and hypothesized hierarchically
organized psychological systems patterned on Jackson's hierarchy of
neural levels [98, p. 488]. This is implied in one of the specifications
of the reflex-arc model, namely, in the sequence of the Systems Un-
conscious, Preconscious, Conscious. Freud's Jacksonian model is closely
related to both the genetic and the topographic models, and its
utility is that it provides the means for coordinating systematically those
behavior phenomena which are not attended by voluntary control
and/ or consciousness with those which are. Not only are the concepts of
the Systems Unconscious, Preconscious, and Conscious (as well as those
of the id, ego, and superego) organized according to this model, but
Freud assumed that every advancement in psychic organization goes
along with a new censorship [117, pp. 122-127], and his conception of
the multiple layering of defenses within the ego also follows the same
pattern [116, 131].

But this does not exhaust the unique significance of this model in
Freud's theory. After all, Janet and Prince based their conceptions on
a similar model, even if Janet did not assume that the "subconscious"
existed under the control of consciousness (in the Jacksonian sense),
but rather that the "subconscious" was created by dissociation caused
by degeneration and precipitated by trauma. In Freud's theory, in-
hibition of lower levels by higher ones served as the model for the
conceptualization of conflict. Thus inhibition became a dynamic event:
the result of a clash of forces. To begin with (1895), these forces were
conceptualized as the libidinal affects vs. the ego, the latter being the
"ruling ideational mass" which serves reality, society, and morality [35,
p. 116]. Later ( 1900), this conception of conflict yielded to that of drives
vs. censorship, the latter representing ego (self -preservative) drives [101,
114]. The final conception (1923) was that of the interstructural con-
flict between the ego and the id, with the participation of the superego
on one or both sides [126, 137]. Thus, the Jacksonian model coordinated
those observations which of old were labeled "conflict" with those from
which "unnoticed conflicts" could be inferred, and it served as the
foundation for the concepts of unconscious conflict, inhibition, un-
conscious drive forces and counterforces, which led to the theory of
symptoms, and ultimately to the theory of mental structure. In
summary: the Jacksonian model served as the foundation for the

The Structure of Psychoanalytic Theory 71

dynamic point of view in Freud's theory, and also contributed to the
development of the topographic and structural points of view.

e. A combined model. Let us now sketch in more detail the entropy
(or economic) model, the behavior sequence restlessness -> sucking on
the breast -» subsidence of restlessness. [For specific references, see Rapa-
port, 267.] Here restlessness is considered the referent of tension accumu-
lation, sucking on the breast that of tension-reducing action on the object,
and subsidence of restlessness that of a state of reduced tension. These
are equated with accumulation of cathexis, action on the cathected
object, and discharge of cathexis, which in turn are referred to drives
reaching threshold intensity, drive action on the drive-cathected object,
and drive-gratification. Finally, the direction implicit in all these se-
quences is conceptualized as the pleasure principle.

This is the primary model of action (conation). It is an action
model because it does not account for thoughts or affects. It is a primary
model because it represents only actions motivated by basic drives, with-
out that intervention of psychic structures, derivative drives, and other
motivations, which is characteristic of most observed actions. It is the
first of the six models to be derived here from the behavior sequence
which is considered to be the model of all motivated behavior. We
shall now derive the primary models of thought and affect, and then
turn to the secondary models.

The primary model of cognition (ideation) was formulated by Freud
[98, pp. 509-510, 533] in 1900: drive reaching threshold intensity-^
absence of drive object -> hallucinatory idea of previous gratification.
When the drive object is absent, drive action is not possible, and a
short cut to hallucinatory gratification takes place. Drive cathexis is
displaced to the memory of past gratifications, bringing these to
hallucinatory intensity. The short cut and direction implicit in this model
were conceptualized by Freud as wish-fulfillment. It is worth noting that
both the pleasure principle and wish-fulfillment (which is its cognitive
equivalent) are abstractions remote from the common-sense meaning
of pleasure and wish. The model extends the economic point of view
to cognitive phenomena, and its concept of wish-fulfillment expresses the
directed, intentional character of cognition. This model makes it possible
to include phenomena like dreams, hallucinations, illusions, daydreams,
reveries in the theory of motivated behavior, and serves as the founda-
tion for those concepts which in the secondary model of cognition co-
ordinate these thought forms with the more familiar cognitive phe-
nomena of ordered veridical thought. It provides the theoretical matrix
for the understanding of free associations and projective techniques, and
concepts for the explanation of the observations in states of need (hunger

72 DAVID RAPAPORT

and thirst [197]), stimulus deprivation [see Hebb's and his associates',
25, 170, 171, 172, and Lilly's experiments, 222, 223], and hypnotic
states [see Gill and Brenman, 33, 148; also Rapaport, 280].

The primary model of affect was formulated by Freud [98, pp. 520-
521] in 1900: drive reaching threshold intensity-* absence of drive
object -» affect discharge. In the absence of the drive object, drive
action not being possible, emergency discharge through affect-discharge
channels takes place. Freud characterized affect discharges first as "sally
gates33 for drive tension [98, p. 520], and later (1911) as discharges into
the interior of the organism (autoplastic adaptation), in contrast to
alterations of external reality by action (alloplastic adaptation) [108, p.
16].21 While other psychological theories postulate direct links between
affective stimuli and the bodily changes and subjective experiences in-
volved in affects, this model — like the topographic one — inserts un-
conscious ideas and drives between affective stimuli and affective re-
sponses. This modification makes possible a unified theory which can
account for anxiety and other persistent affects — for affects which are not
triggered by obvious affective stimuli, as well as for the commonly treated
forms of affect. It also eliminates some of the puzzles confronting, and
various of the contradictions between, the familiar theories of affect
(James-Lange, Cannon, etc.). [For a detailed discussion, see Rapaport,
258.]

The relationship between the primary model of action and the
primary models of cognition and affect is indicated by the presence of
the drive object in the former and its absence in the latter. The rela-
tionship of the cognition model to the affect model is expressed in the
combined primary model of cognition and affect which was formulated
by Freud [116, p. 91; 117, p. Ill] in 1915: drive at threshold inten-
sity -» absence of drive object — > hallucinatory idea and /or affect dis-
charge. It was devised to account for a set of clinical observations.
Clinically, the repressed drive is inferred from its ideational and affect
representations. While in obsessional ideas only the ideational repre-
sentation of the drive is observed (its affect representation usually suc-
cumbing to defense, e.g., repression, isolation, or displacement), in
hysterical attacks only its affect representation is manifest (the ideational
representation succumbing to defense, usually repression). Affect and
idea are thus conceived of as complementary and/or alternative drive
representations.

These primary models unify the traditional trichotomy of conation,
cognition, and affection. They are clearly entropic (economic) models,

21 Of the interactions between environment and organism, those which result
primarily in changes of the organism are called autoplastic, and those which result
primarily in changes of the environment are called alloplastic.

The Structure of Psychoanalytic Theory 73

though they do imply the topographic (reflex-arc) model in their direc-
tion, and the dynamic model in the role played by the drives. They
have adaptive implications in that they posit coordinations, guaranteed
by evolution, between the drive and a real object (the drive object),
and between individuals by means of affect expressions. They have
structural implications: thresholds, which must be reached by drive
intensities before drive action can take place, and drive-discharge
channels. They have genetic implications, in that they are assumed to
pertain primarily to early developmental phases. The secondary models
elaborate these adaptive, structural, and genetic implications.

The secondary model of action was outlined by Freud [98, pp. 533-
534] in 1900: drive reaching threshold intensity ~-> derivative drive
mobilized by basic drive or reaching threshold intensity -> structuralized
delay in the presence of the drive object -> detour-activity searching for,
and means-activity reaching fory the drive object -> satisfaction. Let us
take the relationships between the steps of this sequence one by one.

Drive reaching threshold intensity -> derivative drive mobilized by
basic drive or reaching threshold intensity. This step implies both the
Darwinian and the Jacksonian models: it has both genetic and hier-
archic-structural implications. In the course of development, drives
differentiate into a hierarchy of derivative drives. This drive hierarchy,
in turn, has both adaptive and structural implications. Derivative drives
differentiate according to ontogenetic laws, e.g., that of psychosexual
development, yet the occasions for this differentiation are environmental
(for instance, the periodic unavailability of the drive object, the appear-
ance of substitute objects, the environment's response to and demands
for new ontogenetic achievements, etc. [61, 62, 66]) and the differentia-
tion itself is adaptive. The progressive lessening and change of maternal
and familial care are the environmental counterparts of this adaptive
development and provide the occasions for it. Thus, the development of
derivative drives is not a matter of pure learning, nor is it blindly
regulated by ontogenetic laws.

When the drive object is absent, the drive-discharge threshold is
raised by counter cathexes} and these countercathectic energy distributions
are conceptualized as control and defense*2 structures and derivative
drives.23 This is the outstanding dynamic implication of the model: in-

22 The expression "control and defense" refers to an insufficiently studied set
of phenomena. Certain countercathectic energy distributions effectively prevent
the execution of the motivations against which they are directed: they are termed
defenses. Others merely delay, modulate, and channel motivations: they are
termed controls. In actual observation, instead of this sharp dichotomy, we find a
fluid transition. For a further discussion see [268, part 7].

23 The manifestations of energy distributions are always forces, and clinically,
defenses are always recognized by the appearance of new motivations.

74 DAVID RAPAPORT

hibition, resulting from a conflict between motivating forces and struc-
tures, gives rise to new motivating forces.

A further implication of this first relationship is that action may be
initiated in several ways. The basic drive may initiate it on reaching
threshold intensity. In this case the derivative drives (if they have not
reached threshold intensity) may either be bypassed, or triggered by the
basic drive. But action may also be initiated directly by a derivative
drive which has reached threshold intensity. Or it may be initiated when
an external stimulation provides the excess excitation which brings a
basic drive, or any derivative drive, to threshold intensity. Here again
we encounter the hierarchic arrangement, and the short-circuiting pos-
sibilities discussed in connection with the topographic model.

Now to the second relationship: derivative drive mobilized by basic
drive or reaching threshold intensity -» structuralized delay. This rela-
tionship implies all the above-discussed possibilities for initiating action.
Here the structuralized delay plays, on the one hand, the role played
by the drive-discharge threshold in the primary model: it delays dis-
charge up to a certain point. On the other hand, it plays the same role
as does the absence of the drive object in the primary model: it enforces
delay beyond the point of the original discharge threshold. Structuralized
delay (i.e., control or defense) is conceived as the heightening by counter-
cathexes of the original threshold, so that the object of the drive de-
fended against will be absent (unnoticed or unusable) from the point
of view of psychological reality^ even when present in external reality.
Here the psychological absence of the object plays the same role as its
real absence does in the primary model.

Controls and defenses are conceptualized as structures: their rates
of change are slow in comparison with those of drive-energy accumula-
tions and drive-discharge processes. The delay of discharge, which these
structures make possible, is the crucial distinction between the primary
and the secondary models. In the primary models, the pleasure principle
(the direct discharge tendency) 'prevails; while here, the contrary princi-
ple of least effort — which is one of the referents of the higher-order
concept reality principle (Freud, 1911) — prevails. Threshold and drive
intensity are relative to each other, yet observations necessitate the as-
sumption that the control and defense structures may become relatively
independent of the drives. The relevant observations — for example, the
adaptive role of some behaviors which originated as defensive reaction
formations — are the same as those upon which the concept of autonomy
rests. The structures here are ego structures; their autonomy is one of
the implications of ego autonomy [157, 266, 280], which is akin to All-
port's [8] conception of "functional autonomy."

Now to the next relationship : structuralized delay -> detour-activity

The Structure of Psychoanalytic Theory 75

searching for and means-activity reaching for the drive object. In the
primary model, when the drive object is absent, either the memory of
prior gratification is drive-cathected, resulting in hallucinatory wish-ful-
fillment, or part of the drive cathexis is discharged through the "sally
gate" of affect-discharge channels. In the secondary model, stracturalized
delay postpones discharge, and makes possible detours from the direct
route of gratification and search for the drive object. The concepts of
delay and detour are familiar to psychology (Hunter, 1913; Kohler,
1917). Though Freud introduced them in 1900, psychologists apparently
did not notice that, once taken seriously as concepts, they can account
for the distinction as well as the link between impulsive and controlled
behavior. We will encounter delay and detour again in the secondary
model of cognition.

Finally, the last relationship, detour- and means-activity -» satis-
faction, implies that it is not necessarily drive-gratification that is at-
tained by sequences of this sort.24 We have seen that such sequences may
be initiated by drives or derivative drives, either of which may be
triggered by external excitations. Now — taking the autonomy of the ego
into consideration — we must add that external excitation may also di-
rectly trigger detour- and means-behaviors: the functions subserving
detour- and means-behavior are ego functions. Structuralized delay and
detour, and structure in general (defense-, control-, and means-struc-
tures) are the concepts which enable this theory to account for tension
maintenance and tension increase, and not, as is generally supposed [7],
for tension reduction only. The shift from the "gratification" of the pri-
mary model to the "satisfaction" of this model indicates that full dis-
charge of drive tension gives way to discharge compatible with the
maintenance of tension which is made inevitable by structure formation.

The secondary model of cognition was outlined by Freud [98, pp.
509-510, 533-536] in 1900: drive or derivative drive at threshold in-
tensity -> Structuralized delay -» experiment in thought with small
cathectic amounts to anticipate and plan, locate and act upon the drive
object. The first two steps in this model are the same as those in the
secondary model of action, and the considerations presented above apply.
The relationship between Structuralized delay and experimental action in
thought is the only one to be discussed. According to the primary model
of cognition — which follows the pleasure principle — when drive action
cannot take place, a short cut to hallucinatory gratification occurs,
through the mechanisms of displacement, condensation, substitution,

24 K. Lewin's [216, 220, 221] quasi needs are examples of this. More generally,
the distinction between drive-gratification and satisfaction corresponds to the dis-
tinction between the varieties of action-initiation discussed above on p. 74. For
further discussion see [267].

76 DAVID RAPAPORT

symbolization, etc. In the secondary model, under the same conditions,
structuralized delay prevents the short cut, and detour-behavior involv-
ing ordered thinking results.

The conception of two types of memory organization rests on this
difference between the two models [268, note, pp. 630-631]. In the
drive organization of memories, all the memorial (ideational) representa-
tions of a drive are organized around it and are equivalent to each
other. The syncretic mechanisms enumerated above express this equiv-
alence.25 In the conceptual organization of memories the equivalences
have two determiners: on the one hand, empirical coordinations (fre-
quent contiguity), on the other, logical implications (not all frequent
contiguities are admitted, but only those which are compatible with
logical implications).26

These two memory organizations do not predicate two classes of
thought, but conceptualize two different aspects of any given thought,
with the stipulation that the conceptual organization is hierarchically
higher than the drive organization, and has a controlling function over
it. These cognitive models, and their genetic relation to each other,
represent the first consistent attempt to coordinate, within one theory,
those forms of thought (obsessions, delusions, dreams, etc.) which are
peremptory and those (practical thought, rational thought, rigorously
logical thought) which we can take or leave.

In this secondary model, the intentional, anticipatory potential of
thought derives from the directedness of the primary model, while its
realistic efficacy derives from the structuralized delay which militates
against the immediate discharge and gratification tendency of the pri-
mary model, and thus permits the development and use of conceptual
coordinations.

The secondary model of affect was formulated by Freud [131, chap.
8, particularly pp. 76-79] in 1926: drive or derivative drive at threshold
intensity -» structuralized delay -» affect signal released by the ego from
structurally segregated affect charges. [A more detailed discussion will be
found in "The Psychoanalytic Theory of Affects," 274, and in Organi-
zation and Pathology of Thought, 268.] The first relationship of this
sequence is identical with that of the other secondary models, and the
considerations advanced above apply, with one exception : here the role

25 The term "equivalence" as used here is a generalization of the equivalence
implied in the concept of "equivalent stimuli." It pertains not only to stimuli but
to responses also. It applies not only to "nondiscriminable" stimuli or responses
but also to those whose relationship to each other is that of indicator to indicated.

25 Further discussion of these memory organizations will be found in Organiza-
tion and Pathology of Thought [268] and in "The Psychoanalytic Theory of
Thinking" [265],

The Structure of Psychoanalytic Theory 77

of the drive's reaching threshold Intensity is changed and the con-
sequences of mounting drive tension are anticipated by the ego.

Early in his theory-building, Freud assumed that when no drive
action can take place, the affect-discharge channels serve as "sally gates53
for part of the dammed-up drive cathexes (affect charge). As late as
1915 he assumed [1 17, pp. 109-1 12] that only the drive intensity and the
capacity of the affect-discharge channels determine the affect charge,
and that the latter, before its discharge, is not segregated from the drive
cathexes. This conception still applies to affects in early phases of
ontogenesis.

Later (1926), however, it became necessary to assume that a struc-
tural segregation of affect charges from drive cathexes at large takes
place, parallel with the development of the motivational and structural
hierarchy, as specific affects and affect-discharge channels differentiate at
each level of the hierarchy. Originally, accretion of drive intensity was
assumed to use the affect-discharge channels automatically when drive
action, in the absence of the drive object, is not possible; now, the
absence of objects has been internalized in the form of structuralized
delay, and the ego structures subserving this delay include such as
keep the affect charge segregated and control its discharge also. The
segregated affect charges are therefore under the control of the ego:
when rising drive tension impinges on the ego's defense structures, the
ego uses the segregated affect charge to give an anticipatory affect signal,
which — though of small intensity in comparison to affect discharge —
mobilizes (by virtue of the pleasure principle) countercathexes to rein-
force the defenses, and thus prevents drive discharge [131, pp. 18-20,
112-117]. Affects change in the course of ontogeny from discharge
phenomena into signals, from safety valves for drive tension into antic-
ipations of the means for preventing drive discharge. Under such
"normal" circumstances as bereavement or danger (but also when ex-
posed to wit and drama), as well as under pathological conditions, the
signal affects may yield their place to discharge affects [see 75]. Also,
the segregated affect charge, like all cathectic amounts, may manifest
itself as a motivating force.27 According to the secondary model, affects
may serve as discharge processes, as anticipatory ego-signals for mount-
ing drive tension, and as motivations; thus it unites a wide variety of
observations concerning emotions. Indeed, most of the observations on

ST For instance, actions related to an unconscious sense of guilt may be motivated
by the aggressive impulse which gave rise to the guilt affect, or by a (derivative)
motivation which arose as a reaction-formation to this aggressive impulse, or by
the guilt affect itself which has attained the status of a relatively autonomous
motivation [see 267]. This may be a link to Leeper's [211] motivation theory of
emotions.

78 DAVID RAPAPORT

affects which the various academic theories account for — or fail to ac-
count for — are coordinated in this model.

The behavior forms represented by the secondary models arise
according to ontogenetic laws from those represented in the correspond-
ing primary models, but their development also depends on the en-
vironmental conditions and is thus adaptive.

In contrast to the primary models., all the secondary models involve
structuralized delay, that is to say, progressive, hierarchically layered
structure development. The structures in question are: defense and
control structures, structures which segregate affect charges, and the
means structures which subserve secondary action- and thought-processes.
A parallel development takes place in the hierarchy of motivations : each
step in structure development results in a delay imposed on motivations.,
which in turn gives rise to new derivative motivations and affects. This
multifaceted hierarchic development is the development of the ego
[268, 276] and involves the differentiation of the ego from the id, and
the superego from the ego. The id-ego-superego trichotomy is the
broadest structural articulation of the mental organization and, as such,
a crucial conception of the clinical theory of psychoanalysis. Since it
can be derived from the models discussed, it is not an independent model
and we shall not dwell on it here. [For a similar conception, see Glover,
150.]

The secondary models lean heavily on the Darwinian (genetic) and
the Jacksonian (hierarchic) models, and therefore the structural, genetic,
and adaptive considerations are central to them. Yet they also include
the topographic, economic, and dynamic considerations of the primary
models. Thus, this combined model, which is an elaboration of the
entropic (economic) model, does unite all the models Freud used. But it
does so at the price of falling into six partial models which by their very
nature (if not by that of the theory, or even of the subject matter)
overlap.

4. The comprehensiveness of empirical reference. From the very be-
ginning, the theory implied a comprehensive empirical reference,
though it centered on the psychology of drives and primary processes,
and maintained that its findings concerning these were of unrestricted
validity.28 Freud asserted as late as 1917 [121, pp. 330-333] that the
postponement of the exploration of secondary processes, ego functions,
reality relations, and adaptation was a deliberate policy and not a failure

28 That is to say, they are the ultimate determiners of all behavior. The
conception of ultimate determiner, and the restrictions imposed on it later, will
be discussed further on pp 93 fT., below.

The Structure of Psychoanalytic Theory 79

to recognize their importance. Actually, Freud did take steps toward in-
cluding them in the theory in 1900 [98,' pp. 533, 535] and in 1911 [108],
and consolidated these steps between 1917 [121] and 1926 [131]. But
only in 1939 did Hartmann [157] give the first systematic formulation of
reality relationships and adaptation by expanding the frame of reference
of ego psychology. This systemization, which was accompanied by the
claim that psychoanalysis is a comprehensive system of psychology, was
continued in the studies of Hartmann, Kris, Loewenstein [160, 161, 162,
167, 168, 206], Rapaport [268, especially part 7; 267, 277], Jacobson
[190, 191, 192], and Gill and Rapaport [149].

In the meanwhile, and even before these developments in the main-
stream of psychoanalysis, adaptation and reality relationships, especially
the role of interpersonal relations and society, were central to the theories
of Adler, Horney, Sullivan, and Kardiner [see Munroe, 240]. Erikson
[56, 57, 58, 59, 60," 61, 62, 63, 66] was the first to unite this tributary of
theoretical development, which enlarged the actual realm of empirical
reference, with the mainstream of the theory.29

Finally, in the late thirties, forties, and fifties, the influence of psy-
choanalysis and of the new psychoanalytic ego psychology expanded to
the whole of psychology, first through projective techniques into clinical
psychology, then into experimental clinical psychology, and finally into
experimental psychology proper [see 309]. Thus the original claim of
comprehensiveness for this theory is gradually being realized.

If we must single out an outstanding limitation of this theory's claim
to comprehensiveness, then we should choose its lack of a specific learning
theory. Psychoanalysis has created grounds on which contemporary
learning theories (Hull, Bollard, Miller, Mowrer, etc.) can be sharply
criticized, and its conception of the primary process (e.g., the drive
organization of memories) and of the secondary process (e.g., the con-
ceptual organization of memories) can be regarded as foundations for a
theory of learning. But like Lewinian and Gestalt psychology, it has failed
to offer a specific alternative learning theory. Though Hartmann's [157]
automatization concept seems to open a new approach to the problem
of learning — as did K. Lewin's [221] ossification concept — so far no one
has used it. The problem of learning — how a process turns into a
structure, or in other words, the long-term survival and availability of
experience — has not been solved by psychoanalysis either.

5. Quantification and mensuration. Psychoanalytic theory does con-
tain quantitative considerations (particularly in its economic point of
view), but the translation of these into actual measurements presents

29 Hartmann, whose work is an indispensable link between Erikson's work and
classic psychoanalytic metapsychology, laid the metapsychological groundwork
for this unification but did not actually undertake it.

80 DAVID RAPAPORT

difficulties which have not been overcome. Some of these difficulties will
be mentioned here and in Section V. below.

Freud did not take a theoretical stand implying unquantifiability;30
yet neither he nor any other psychoanalyst tried to quantify the variables
of the theory. However useful and indispensable the theory is clinically,
however much light it sheds on a broad range of human phenomena, and
however consistently everyday and clinical experience confirms its help-
fulness, as a theory it requires exact tests of confirmation which in turn
require the mathematization of the relationships posited by it. The
obstacles to mathematization are: (a) The basic independent variable
(drive cathexis in general, libido in particular) postulated by the psycho-
analytic theory is an intrapsychic one, related to organic changes and
intrapsychic structural conditions, rather than to external stimuli; thus
it is hard to manipulate31 and measure. ( b ) The avenues through which
such variables may exert their causal effect are multiple and interchange-
able (cf. Tolman's vicarious function, Lewin's substitute tasks, and
Heider's equifinality), and thus hard to predict, observe, and measure.32
(c) The distance between the theory's major variables and the observed
phenomena makes it uncertain whether or not any measure obtained
actually quantifies a particular variable.

But these obstacles need not prevent mathematization (e.g., quanti-

30 We do know, however, that he occasionally took a practical stand to the
effect that the theory needs no experimental confirmation [cf. Rosenzweig, 293].
I am indebted to Dr. Saul Rosenzweig for a personal communication which indi-
cates that Freud, in a conversation with H. A. Murray and in one with R. Grinker,
seems to have retracted this stand.

31 Not the least of the obstacles to manipulation is the inviolable privacy of the
subjects.

32 The point is frequently made that Freud's failure to quantify his variables
was due to his having come from "another tradition," and that the continued
avoidance of quantification stems from the development of psychoanalysis "apart
from academic psychology." True, Freud's neurological research was in the area
of nonquantitative neuroanatomy. It is also possible that, as Holt [177] suggests,
Freud's experience with Fliess's "numbers game" — combined with a general
limitation in mathematical thinking, which he mentions repeatedly [94] — made
him averse to quantitative considerations. Still, these arguments seem to miss the
mark, and distract attention from the lack of quantitative methods applicable to
intrapsychic variables. Academic psychology has only recently begun to be con-
cerned with such methods. In addition, these arguments disregard that the Helm-
holtz tradition was the matrix of both Freudian and academic psychology, and that
the biology of that time was not centered on quantification but rather on the
significance of the single case and on the tracing of genetic connections. The theory
of evolution seems to have been a "good science" even though the complex statistics
applied to it by G. G. Simpson [311, 312] were not available to Darwin. In just
what sense Darwin's and Freud's theories are good sciences is an interesting but so
far — to my knowledge — unanswered question.

The Structure of Psychoanalytic Theory 81

fication), though they do suggest that the road to It will be long and
arduous. It is doubtful that the long hierarchic chain of intermediary con-
cepts interposed between the major explanatory constructs and the ob-
servables can be bypassed, and that direct relations can be found between
them. This highlights the importance of theory construction, since only
a tightly built theory (with clearly stated definitions and implicative
rules) can support confirming tests on observables which are at a great
distance from the constructs: the models discussed above show that in
this theory the implicative rules change with the distance from the basic
variables.

Many features of observables can be counted, rated, and measured,
but the observables alone cannot tell us which features and what method
of counting or measuring them will reveal the relationship between them
and the explanatory constructs: only theory can do that. A certain
amount of trial-and-error (ad hoc] quantification is inevitable, but it
will never yield a theory: theory is the product of theory-making. The
confirmation or refutation of a theory requires that we quantify those
features of the observables which correspond to the dimensions of the
theory's variables. The ad hoc quantifications are not necessarily useless;
they may be the means by which the deductions from the theory and the
inductions from the observables are brought progressively closer to each
other, and by which the essential measurables are progressively selected
from the multitude of all measurables. But this selection cannot be
achieved by blind measurement unguided by theory: there is no end to
that.33

The first steps toward quantification are (a) systematic mastery of
the theory as it exists at present, (b] systematic attempts to tighten the
theory, (c] the selection of measurables relevant to the variables of the
theory. So far, no attempt at quantification has included these steps.
Most of the experimenters who have attempted to confirm or refute the
relationships posited by psychoanalytic theory were unaware of the
nature of, and the variables involved in, the relationships which they set
out to test.

This may seem to be a sad picture of the theory and a summary
indictment of the experimenters who have tried to deal with it. Neither
of these is intended. We are blinded by the rapid development of new
sciences in our time. The rapid growth of biochemistry and biophysics
was possible because they had the solid foundations of several thousand
years of physics and chemistry. Some psychologists are bent on linking
psychology to those sciences now, hoping for an equally spectacular
growth of psychology. Others are more patient. They do not deplore the
present state of the theory, nor consider the experimenters to be fools.

33 For similar considerations in geology, see Rich [287].

82 DAVID RAPAPORT

In their eyes these difficulties are phenomena of a very early phase in the
development of a science. Clinical observation shows that conscious in-
formation does not eliminate symptoms rooted in unconscious forces and
that conscious intent is usually no substitute for the lack of unconscious
motivation. Likewise, we may assume that consciously borrowed method-
ological sophistication, however much it may help otherwise in develop-
ing psychology, cannot circumvent the long and time-consuming process
all sciences have gone through. The process of development which brings
about the interplay between the observables and the theories is always
slow.34 Quantification and methodological sophistication are late prod-
ucts of any science and as such they should be long-range goals: mis-
taking them for proximal goals can render a science impotent.

6. Formal organization. The expositions of psychoanalytic theory
have been informal rather than systematic; in the main they were di-
rected by internal consistency within the theory and between observables
and the theory. In the last twenty years attempts at systematic formula-
tion [26, 73, 150, 157, 166, 167, 267, 268, 274] have been made, but no
hypothetico-deductive system-building is in sight.

II. THE STRUCTURE OF THE SYSTEM

In order to discuss the systematic independent, intervening, and
dependent variables of the psychoanalytic theory, it seems necessary to
sketch the theory's structure.

A. The Subject Matter of Psychoanalysis Is Behavior (the Empiri-
cal Point of View)

This proposition has often been overlooked, probably because the
theory's stress on unconscious processes and drives, psychological struc-
tures, dynamics and economics obscured the fact that it conceives of all
of these as explanatory concepts of behavior.

34 We have some idea why this process is so slow. If logic, methodology, and
mathematics were the pacemakers of development in sciences, this development
could be fast enough in psychology. But the pacemaker is not methodology — it is
human invention. ("Developmental projects," "crash programs," and "inter-
disciplinary teams" are effective only in highly developed sciences or else in
situations where the makeshifts of pooled ignorance are the most that can be
had.) Methodology, since it deals with relationships of concepts, all of which are
potentially valid, can go on continuously, building ever-new "castles in Spain." But
human invention consists of discontinuous events, each of which requires long
preparation, since in it an individual's thought patterns must come to grips with
patterns of nature, and only those rare encounters in which a unique human
thought pattern actually matches a unique pattern of nature will matter. If the
match is not specific and precise, or if the individual is not prepared to recognize
it, or if he does recognize it but is not ready to use it, the moment is lost.

The Structure of Psychoanalytic Theory 83

Behavior in this theory is broadly defined, and includes feeling and
thought as well as overt behavior, "normal" as well as "pathological/3
frequent as well as unique forms of behavior. This corollary too has
often been overlooked, probably because of the stress in psychoanalytic
literature on "latent behavior" and on pathology, both of which served
as points of departure for the theory. Indeed, not before Hartmann's
[157] major study (1939) was it directly stated that psychoanalysis is a
general psychology which embraces the study of normal as well as
pathological behavior,35 though the principle of the thoroughgoing psy-
chological determination of all behavior has been the cornerstone of the
psychoanalytic theory from the beginning and was explicitly stated in
1905 [99].

Thus, all appearances to the contrary notwithstanding, psychoanalysis
does not differ from other theories in its view of the subject matter of
psychology (though it defines behavior far more comprehensively than
most), nor in its assumption of determinism (though it probably de-
manded this earlier and in a more sweeping fashion). Yet it does differ
from other psychologies in assuming psychological determinism, and in
its stress on "latent behavior" in general and on the unconscious deter-
minants of behavior in particular (cf. Section II. E., below).

B. Behavior Is Integrated and Indivisible: The Concepts Con-
structed for Its Explanation Pertain to Different Components of Be-
havior and Not to Different Behaviors (the Gestalt Point of View)

In the clinical parlance (and even in the theoretical writings) of psy-
choanalysis, the explanatory concepts are anthropomorphized, reified, or
at best presented in existential terms, giving the impression that they
refer to entities or at least that each of them refers to a specific behavior.
But this is not consistent with the theory. The tendency to anthropo-
morphize and reify, and the preference for hypothetical constructs prob-
ably derives from clinical practice, where there is a premium on the
"plausibility" and "uncomplicated everyday application" of concepts.

In concrete terms: no behavior can be described as an id behavior,
or an ego behavior, or a conscious behavior. These concepts all refer to
specific aspects of behaviors and not to specific behaviors. Every behavior
has conscious, unconscious, ego, id, superego, reality, etc., components.
In other words, all behavior is multiply determined (overdetermination).
Since behavior is always multifaceted (and even the apparent absence
of certain facets of it requires explanation), the conception of multiple

35 It is noteworthy, though, that Freud's recently discovered manuscript [94,
appendix], which is the predecessor of the theory of psychoanalysis, has the scope
of a general psychology.

84 DAVID RAPAPORT

determination (or overdetermination) may be regarded as a purely
formal consequence of this method of conceptualization. This naturally
does not rule out the possibility that the conception of overdetermination
is also required by the nature of the observations; in fact overdetermina-
tion as a concept was originally introduced [35, pp. 156, 219] in refer-
ence to observations,36 rather than on purely theoretical grounds. From
the very beginnings of psychoanalytic theory, observations made the con-
cept of overdetermination both necessary and central. Academic psy-
chologies did not develop such a concept, probably because their methods
of investigation tend to exclude rather than to reveal multiple determina-
tion. But they did not escape the problem itself: every behavior phe-
nomenon has perceptual, learning (memorial), conceptual (cognitive),
motor, etc., components; and the rival psychological theories (perceptual
theory of cognition, learning theory of perception, motor theory of
thought, etc. ) show both the presence of the problem and the confusion
resulting from a failure to face it squarely.37

36 For example, when a subject executes the posthypnotic suggestion to shut a
door and explains that he did so because of the draft [53], then his action is de-
termined both by the hypnotic suggestion of which he is not conscious and by
his conscious intention to escape the draft. Dr. A. B. Wheelis (San Francisco)
suggests (personal communication) that there are distinctions among overdeter-
mination, multiple determination, and multiple levels of analysis, which hinge
on whether the determiners are independent and sufficient causes of the behavior
in question (overdetermination) or not (multiple determination). It appears,
however, that in psychoanalytic theory neither such independence nor such suffi-
ciency of causes can be demonstrated or perhaps even defined. The fact that to
escape the draft would be, under other conditions, a sufficient cause for shutting
the door does not make it a sufficient cause in the posthypnotic situation. The
matter of the "independence" of causes is an autonomy issue (cf. Section II. H.,
below). Overdetermination, to my mind, implies precisely such a lack of inde-
pendence and sufficiency of causes and is inseparably connected with the multiple
levels of analysis necessitated by this state of affairs. Mr. J. Zsoldos (Maabaroth,
Israel) suggests (personal communication) that the "overdetermination" issue
crops up where "weak (sensitive) systems'5 are exposed to overwhelmingly large
forces, that under such conditions simple functional relationships do not obtain,
and quantitative analysis is possible only in terms of statistics; so that "weak sys-
tems" have only statistics, not "laws." This suggestion seems to imply that the
overdetermination issue is the psychological counterpart of the controversy between
Einstein's theory and present-day atomic physics. The psychoanalytic theory of
overdetermination as it stands — if I read it correctly — implies laws and not statis-
tics. To use Einstein's phrase, "The good Lord does not play dice" in this theory
either. Nevertheless, the possibility of a statistical interpretation of overdetermina-
tion must be kept open, even if reluctantly; a specific and workable statistical
interpretation would be preferable to an interpretation which assumes the existence
of laws but does not specify any implicative rules and thus permits neither con-
firmation nor refutation.

37 "Field theories" may be looked upon as attempts to meet this problem.

The Structure of Psychoanalytic Theory 85

This psychoanalytic proposition has implications over and above
that indivisibility of behavior from which the considerations pertaining
to overdetermination stem. It requires that each of the conceptually
differentiated aspects of behavior, as well as the spatial and temporal
context of behavior, be treated as an integrated whole. But we need not
pursue this point further: it seems to coincide grossly with the general
postulate of Gestalt psychology.

C. No Behavior Stands in Isolation: All Behavior Is That of the
Integral and Indivisible Personality (the Organismic Point of View)

This thesis demands that the explanation of any behavior fit into the
theory of the workings of the total personality.38 Freud's most direct
statement of this thesis is probably that pertaining to dreams. Once he
had developed the theory of dreams, he raised the question: what kind
of theory of personality could embody this dream theory. In Chapter 7
of The Interpretation of Dreams he proceeded to construct the frame-
work of such a theory of personality [98, pp. 469, 470, 485-486].

Yet this implication of psychoanalytic theory, too, has been overlooked
by many psychoanalysts and psychologists, probably because the stress
on the central role of drives made it appear to the psychoanalyst that the
fundamental drives sufficiently guarantee the unity of behavior and per-
sonality, and gave the psychologist the impression that in this theory the
"atomistically" conceived behavior fragments are held together only by
the "glue" of the drive concept. The organizing, integrative role of the
secondary process (1900), however, speaks eloquently against both of
these views [98, pp. 533-536], and the "structural point of view" (ego,
id, superego, etc.), which clearly embodies principles of cohesiveness
other than drives [126, pp. 15-18; see also Nunberg, 245, and Erikson,
63], should have dispelled these misconceptions. It did not.

To be sure, it is easy to find passages in Freud which lend them-
selves to atomistic interpretation, but the theory itself does not. French
[87, 88, 89] devoted his major work to demonstrating the role of the
integrative field in psychoanalytic considerations. Psychoanalytic studies
in psychosomatics embraced Goldstein's organismic view as "consistent"

38 It may be objected that Freud did not explicitly formulate the organismic
point of view and that only the organismic biologists and Wertheimer, Goldstein,
and Wheeler arrived at it. But our task here is not limited to a collation of Freud's
explicit systematic formulations. Recognition of the organismic thesis of psycho-
analysis is the more important, since Gestalt as well as personalistic psychologists
viewed psychoanalysis as an atomistic and mechanistic theory. Wertheimer was
vehement about this in his lectures and conversation, and G. Allport [7] outspoken
in his writings. The attitude of many practicing psychoanalysts in regard to sym-
bols, dream interpretations, etc., to the contrary notwithstanding, this thesis ap-
pears to be a basic implication of Freud's theory.

86 DAVID RAPAPORT

with psychoanalysis. The projective techniques, which developed under
the impact of psychoanalytic conceptions, borrowed from psychoanalysis
the projective postulate [259, 261] that all behavior is integral to, and
characteristic of, the behaving personality. Nevertheless, this implica-
tion of psychoanalysis remained so remote that French [85] in 1933
(before his familiarity with K. Lewin), and later Mowrer [239], as well
as Bollard and Miller [48], found it feasible to link psychoanalysis to the
atomistic conditioned-response theory of learning.

What this organismic point of view asserts is not that each behavior
is a microcosm which reflects the macrocosm of the personality, but
rather that an explanation of behavior, in order to have any claim to
completeness, must specify its place within the functional and structural
framework of the total personality and, therefore, must include state-
ments about the degree and kind of involvement, in the behavior in
question, of all the relevant conceptualized aspects of personality.

D. All Behavior Is Part of a Genetic Series, and through Its Ante-
cedents, Part of the Temporal Sequences Which Brought About the
Present Form of the Personality (the Genetic Point of View)

This thesis implies that every behavior is an epigenetic product [58]
and thus can and must be studied genetically for its full explanation
[166]. However, it implies neither a view of behavior as the "matura-
tion53 of a preformed behavior repertory, nor one according to which be-
haviors "develop" from accumulating experience; rather, it views be-
havior as the product of an epigenetic course which is regulated both by
inherent laws of the organism and by cumulative experience.

The genetic point of view does not conflict with K. Lewin's in-
sistence that only forces and conditions which are here and now present
can in the here-and-now exert an effect;39 it asserts simply that much of
what "exists" here and now in the subject can only be known through a
genetic exploration of its antecedents. This implies that descriptively
identical behaviors may differ in their psychological significance, depend-
ing on their genetic roots. But it also implies that the empirical relevance
of a behavior to a situation in which it occurs alone does not necessarily
explain it and that the explanation must also take into consideration
the epigenetic laws which brought the behavior about. Indeed, it is
peculiar that it should have been Lewin who criticized the genetic point
of view, when he more than any other psychologist stressed the distinc-
tion between genotype and phenotype and sharply criticized the use of
achievement concepts. He gave the example : identical typewriting speeds

39 Nor does it clash with Lewin's [218] and Chein's [43] point that the past
reconstructed by the patient in psychoanalysis is the past as he views it in the
present.

The Structure of Psychoanalytic Theory 87

of applicants for a job provide insufficient information, since they may
be products of maximal exertion or routine approach, disuse or peak
efficiency, recent training or established working level [220, pp. 89-91].
We must conclude that without the exploration of its genetic antecedents,
a behavior can only be described in terms of achievement concepts.

The genetic point of view refers to the history of the drive proc-
esses which express themselves in a given behavior, to the history of
the structures (e.g., those corresponding to "abilities") used in it, and to
the history of the subject's relation to the situation in which the be-
havior occurs. An example of the distinctions implied here: a sudden
attack of stammering, which is brought about by defense against an
aggressive impulse. A genetic exploration will take into consideration
those past experiences, and the controlling structures crystallized from
them, which modulated the development of the aggressive drive and thus
gave the power to arouse aggression to situations like the one which
aroused the subject's anger in this instance. It will also consider the past
experiences which led to defense against aggression in general or against
that particular kind of aggression which came into play in the given
situation. It will extend to those past experiences which made the verbal
avenue for the expression of anger particularly vulnerable to defense and
to those past events which shaped the stammer, that is, the form which
defense took in this situation. In each of these instances, experience
denotes both the historical event in its external setting and the internal
situation of the subject, including the specific phase of his maturation
and development.

Although the genetic point of view does not refer specifically to the
contextual (spatial-temporal) determination of behavior, it does imply
contextual determination. Moreover, it does specifically refer to the intra-
psychic context : to the contemporary state of the personality as a whole
and as a genetic product.

The genetic character of the psychoanalytic theory is ubiquitous in
its literature. The concept of "complementary series"40 is probably the
clearest expression of it: each behavior is part of a historical sequence
shaped both by epigenetic laws and experience [101, summary; 279];
each step in this sequence contributed to the shaping of the behavior
and has dynamic, economic, structural, and contextual-adaptive relation-
ships to it. Such complementary series do not constitute an "infinite
regress": they lead back to a historical situation in which a particular
solution of a drive demand was first achieved, or a particular apparatus
was first put to a certain kind of use [cf. 166].

But this formulation is incomplete because it disregards those ob-
servations to which the concept of autonomy refers [157, 161, 280]. Cer-

* Erganzungsreike, see Freud [121, lecture 21] and Fenichel [73, pp. 121ff.].

88 DAVID RAPAPORT

tain behaviors do (all tend to) cease to be shaped further by their re-
currence: they become automatized [157] and relatively autonomous
from their genetic roots; they take on a tool- or means-character and at-
tain a high degree of stability. However, automatization and autonomy
make not only for stability, but also for the availability of the behavior as
a means of adaptive performance. These automatized behaviors can also
be studied genetically, but from that point at which they become auto-
matized, their "complementary series" proves relatively unrevealing,
since from there on situation and context may bring them into action,
even in the absence of the motivations which gave rise to them originally.
In Section II. H., pp. 93-97, we return to the concepts of automatization
and autonomy. They are akin to Woodworth's [329, pp. lOOff.] concept
of "habits as drives," and to Allport's [6; 8, pp. 76ff.] concept of func-
tional autonomy, but they are more specific and more differentiated than
either of these.

Psychoanalysis as a genetic psychology deals with the genetic roots
of behaviors, with the degree of autonomy behaviors attain, and with the
genetic roots of the subject's relation to the reality conditions which
codetermine the appearance of a behavior at a given point in the person's
life. Yet the first formal statement of the genetic point of view of psy-
choanalysis is that of Gill and Rapaport [149].

E. The Crucial Determinants of Behaviors Are Unconscious (the
Topographic Point of View)

Per se, this thesis is not alien to any psychology, and particularly not
to those psychologies which exclude all phenomena of consciousness from
their subject matter, and thus have to assume that the determinants of
behavior are extraconscious. All psychologies deal with conditions "un-
noticed" by the subject, and with "unnoticed" or "unnoticeable" proc-
esses underlying his behavior. The psychoanalytic thesis of unconscious
determination, however, differs from these [98, pp. 543-544; 110] in
several respects: (1) it explicitly conceptualizes that which is unnoticed
or unnoticeable [110]; (2) it asserts that the unnoticed or unnoticeable
can be inferred from that which is noticed by the subject (and/or the
observer), by means of the effects of the unnoticed and/or the unnotice-
able upon that which is noticed [99] ; (3) it asserts that the rules govern-
ing the noticed are different from those governing the unnoticed, and
that the unnoticed can be inferred by considering the deviations of the
noticed from its usual patterns [117, pp. 118-122]; (4) it makes a
systematic distinction between the unnoticed and the unnoticeable (the
unnoticed can become conscious, whereas the unnoticeable, by definition,
cannot) ; it expresses this distinction by the terms "descriptive" vs,
"dynamic" unconscious, and conceptualizes it as the distinction between

The Structure of Psychoanalytic Theory 89

the Systems Pr^conscious and t/rcconsclous [98, pp. 543-544]; (5) while
other psychologies treat the unnoticeable in nonpsychological terms (brain
fields, neural connections, etc.), psychoanalysis consistently treats it in the
psychological terms of motivations, affects, thoughts, etc.

The differences between the laws governing the Conscious and those
governing the Unconscious are expressed in the concept of primary and
secondary processes.41 In the early phases of the theory, the Cs and the
Ucs were considered systems of paramount significance. Later on ( 1923)
they were subordinated to the structural conception id-ego-superego
[126], and still later (1938) were relegated to the role of "qualities"
[140].42 In keeping with Freud's early formulations [35, 98], recent
contributions to ego psychology treat consciousness as a superordinate
sense organ. They attribute to it a complex hierarchic layering (states
of consciousness) [268, 270, 276], and thus treat it on a level of abstrac-
tion different from that accorded the unconscious.

F. All Behavior Is Ultimately Drive Determined (the Dynamic
Point of View)

This thesis of psychoanalysis has become only too well known in
a doubly distorted form : all behavior is determined by sex. The qualifier
"ultimately" was omitted, and sex, libido, drive, and psychosexuality
were equated. It is certainly true that, until recently, the drives most
closely studied by psychoanalysis were the sexual drive and its partial
drives. But psychosexuality was defined in such a broad way that it was
by no means synonymous with "sex" [101]. Self -preservative and ego in-
stincts were also discussed early [101, 114, 115, 121], but were dropped
later since they did not prove helpful in organizing empirical evidence.
The history of the theory of drives (narcissism, instinctual vicissitudes,
life and death instincts, monistic drive theory, aggressive drive) suggests
that the early centering on libidinal drives helped Freud to explore the
nature of drives and their motivational role [115], but did not settle the
theory of drives itself [26]. In spite of some recent advances [160, 168],
it is still unclear how many and what kinds of drives need to be
postulated.

The crucial role attributed to libidinal drives is not a theoretical
necessity in this system. It seems to derive from two of Freud's major
achievements : the conception of the determination of behavior by drives
and the observation of infantile sexuality. The fact that the theory linked
these two to each other very early may have retarded a full assessment
of the role of libidinal drives in psychological life.

41 The relationship of Conscious vs. Unconscious to primary vs. secondary proc-
esses is not, however, a one-to-one coordination.

42 Section I. B. 3. a. presented the issues discussed here from another angle.

90 DAVID RAPAPORT

Infantile sexuality was an empirical discovery made by the method
of psychoanalytic reconstruction and verified by the method of direct ob-
servation on infants and children. Though of empirical origin, it grew
to systematic importance because it provided the example for the
development of drives [101].

The broader principle of drive determination was an empirical as
well as conceptual discovery. The empirical discovery embodies, besides
its novel aspect, two familar observations: (1) behavior is not always
triggered by external stimulation but often occurs without it, as though
spontaneously; (2) behavior (which by scientific fiat is causally deter-
mined) evinces a goal-directedness, a purposive, teleological character.
The conceptual discovery, which took the form of the definition of the
drive concept, was the first large-scale attempt to cope with both
of these observations simultaneously. The drive is defined as a causal
agent inherent in the organism [115, p. 64], and thus it can account
for the apparent "spontaneity" of behavior. Moreover, since the definition
makes the effectiveness of the drive dependent on an environmental
condition, namely, the presence of the drive object, it can also account
for the purposiveness of behavior. This coordination of drive and drive
object — which is assumed to be guaranteed by evolution — at first
tolerates little if any means-activity and demands immediate consum-
mation (pleasure principle). In the course of development, it becomes
more flexible, and permits delay and interpolation of means-activities,
though it selects and organizes these in the service of consummation.
Later on it permits substitute goals and a variety of means- and con-
summatory-activities, until finally it prescribes only the consummatory
behavior, and provides no more than the motivational framework for in-
strumental behavior. This conception of motivation accounts not only
for the spontaneity and teleology of behavior, but also for behavior
elicited by external stimulations, since the latter may be conceived of as
drive objects, or substitutes for them.

If psychological theories can be divided into two classes according
to whether they consider the human psyche a tabula rasa on which ex-
perience writes, or an organization of actualities and potentialities which
limits and regulates the extent and kind of changes that experience can
bring about, then the drive conception certainly belongs to the latter
class. In philosophical systems this distinction is crudely paralleled by
Hume vs. Kant [cf. 9, pp. 7ff.; 252, 254, 262]. In terms of psychology
the distinction is paralleled, for instance, by the conception of passive
registration of experience vs. active organization of it, a distinction which
involves the nurture-nature controversy. Psychoanalysis was one of the
first theories to recognize the interaction of nature and nurture in the
development of behavior. Drives represent the "nature" factor; and

The Structure of Psychoanalytic Theory 91

their vicissitudes, in the course of experience, the interweaving of nature
and nurture. Moreover, the coordination of drive and drive object ex-
presses a primary coordination given by evolution between human nature
and its environment and is thus a psychological representation of the
biological adaptedness of the species to its environmental, ecological
niche.

Finally we come to the "ultimate determination of behavior by
drives." Here we meet relationships like those connected with the
"genetic point of view." While early psychoanalysis actually maintained,
without reservation, the thesis of "ultimate drive determination," the
increasing evidence for the "indivisibility of behavior" led to the realiza-
tion that behavior, in so far as it can be said to be determined by drives,
must also be said to be determined by defenses and/or controls. More-
over, with the development of ego psychology, the question was no longer
which of these was the ultimate determiner of behavior but rather in
what respect and to what extent was each the determiner of a given be-
havior [cf. Waelder, 320]. Finally, behaviors were encountered in which
drive determination was in abeyance. This led to the concept of ego
autonomy (cf. Section II. H.).

Thus the thesis of the ultimate drive determination of behavior,
while it remains valid in psychoanalysis, must be regarded in the con-
text of the other theses here discussed, which qualify it and limit its
scope. The concepts of drive fusion, drive differentiation into partial
drives, conflict, etc., all pertain to the dynamic point of view and in-
dicate limitations to the conception of ultimate drive determination.

G. All Behavior Disposes of and Is Regulated by Psychological
Energy (the Economic Point of View)

This thesis, too, has a history. In the first phase of psychoanalytic
theory (abreaction theory — up to 1898), psychological energy was
equated with affects, and the "defenses" which prevented abreaction were
not conceptualized in economic terms [35, 98]. In the second phase
(1900-1926), psychological energy was conceptualized as drive energy,
and the methods used in discharging it as the primary process. It was
recognized that other (secondary) processes, using minute quantities of
energy, exert a regulative function over those which dispose of drive
energies [98, particularly chap. 7; 108, 116, 117, 119, 120]. The
relationship between these two kinds of processes was conceived much
like that described nowadays as obtaining between power engineering
and information engineering [cf. Wiener, 326, pp. 53-56; Rapaport,
264], In this phase, however, little attention was paid to the nature and
origin of the secondary process. In the third phase (after 1926), some

92 DAVID RAPAPORT

understanding was gained of the secondary process, of its congenital roots,
and of the progressive ontogenetic transition to it from the primary
process [131, chap. 8, especially pp. 82, 112-117; cf. also 274].

All behaviors have both primary process and secondary process
aspects, though one or the other may predominate. The primary process
operates with drive energies, and its regulative principle is the tendency
toward tension reduction (pleasure principle) :43 it strives toward imme-
diate discharge of energy accumulations, by a direct route and by means
of the mechanisms of displacement, condensation, substitute formation,
symbolization. The secondary process operates by the principle of least
action, is oriented toward objective reality, and finds, through delays
and detours, by experimental action in thought, the safest course toward
the sought-for object in reality, suspending the discharge of drive energies
until the object has been found [98, pp. 533-536; 108].

In the course of development, hierarchically layered structures arise
(defenses and controls) which act as "dikes." These not only delay or
prevent discharge, but also dimmish the drives5 tendency toward im-
mediate discharge. These structures are conceived of as built by "bind-
ing" drive energies to heighten the originally given drive-discharge
thresholds [98, pp. 533-534; 116, 117]. Their effect of diminishing the
drives' tendency toward immediate discharge is conceptualized as
"neutralization," special instances of which are referred to as delibidiniza-
tion, deaggressivization, or sublimation [126, pp. 61-65; 206, 164].
These processes of binding and neutralization make cathexes (hyper-
cathexes, attention-cathexes) available to the secondary process, to be
used in small quantities for experimental action in thought [108, p. 16],
as signals in the form of affects [98, p. 536; and 131], and as counter-
cathexes (against drives) for building new and for reinforcing existing
defensive structures. Once a process of "neutralization" is assumed, the
original dichotomy of primary and secondary processes yields to a
hierarchic model in which these two represent theoretical extremes and
the actually observed phenomena represent intermediary forms [206;
268, e.g., p. 536]. The energies of lesser degrees of neutralization (drive
derivatives) show characteristics of their drive origin^ whereas those of
higher degrees do not, and are at the disposal of the ego.44 Sources of

43 We will not discuss here the distinction between an "optimal35 and a "maxi-
mal" lowering of tension. The latter has been assigned to the so-called "nirvana
principle" (associated with the "death instinct") — a speculative excursion which
does not seem to be an integral part of the theory. Cf. Freud [123], Fenichel [74],
and Hartmann, Kris, and Loewenstein [168].

44 While the ego builds new or employs existing defenses against drive deriva-
tives of low neutralization, it can make use of derivatives of high neutraliza-
tion, since it can deal with these by means of its controls. This difference seems to
be akin to the difference between all-or-none vs. graduated processes.

The Structure of Psychoanalytic Theory 93

neutralized energy other than drives are also compatible with the theory
[161, pp. 86-87; 162, p. 21].

These psychological energies are not equated with any known kind
of biochemical energy. They do not correspond to the muscular energy
expended in overt behavior. The differences in the quality (mobility vs.
degree of neutralization) of psychological energy45 correspond to the
observed differences between overvalent thoughts (e.g., obsessions) and
passing thoughts (e.g., logical thinking), between impulsive or com-
pulsory actions and actions which are a matter of choice. There is an
obvious conceptual similarity between Freud's energies and Lewin's
tensions and between Freud's drives and drive objects and Lewin's forces
and object valences, though there are also significant differences be-
tween them. They are alike in that they cannot be expressed in the
mathematical formulas in which physics expresses its energy concepts,
yet they are referents of phenomena which seem to abide by the laws
of energy exchanges — conservation, entropy,46 least action. (However,
it is neither implied nor ruled out that biochemical energy exchanges
may eventually be discovered which correspond to the exchanges of
psychological energy inferred from behavior by psychoanalysis.) They
differ, among other things, in that Lewin's concepts do not account for
the differences in quality here discussed.

From the point of view of the energy economy of the organism, the
exchanges of psychological energy may be considered as the work of an
information engineering network which controls the biochemical energy
output of overt behavior. But this network itself is multiply layered, so
that ever smaller quantities of energy control the networks which carry
and dispose of greater quantites of energy [cf. Wiener, 326]. For further
discussions pertinent to the economic point of view, see Sections I. B.
3. b. andV.A.

H. All Behavior Has Structural Determiners (the Structural Point

of View)

The simplest way to put the issue of structure is to point out
that drive energies can be conceived of only within well-defined systems
which have definite thresholds of discharge. The structural conception
could well have been first necessitated by the observations pertaining
to such discharge thresholds; and the prototype of the conflict between

45 Speaking about the "quality" of energies does not contradict the fact that
energy is a quantitative construct. Physics too speaks of different kinds of energies:
heat, light, etc. But in psychology we do not yet have transformation equations
to express the quantitative relationships of these qualities to each other.

46 The validity of the economic point of view is unaffected by Bertalanffy's
conception of "open systems" [20].

94 DAVID RAPAPORT

drives and structures could have been the relationship between drives
and thresholds. Actually., however, this was not the origin of the struc-
tural point of view, even though the early (1895) neuropsychological
form of the theory would allow such an interpretation [94, appendix],
Recently (1939) Hartmann [157; see also 162] pointed out that not only
thresholds but ego apparatuses like memory, perception, and motility are
also structural givens.47 But the structural point of view did not originate
in these structural givens either.

It was observed that drives do not unequivocally determine behavior
in general, nor symptom formation in particular. In contrast to the drive
processes, whose rate of change is fast and whose course is paroxysmal,
the factors which conflict with them and codetermine behavior appeared
invariant, or at least of a slower rate of change. The observation of
these relatively abiding determiners of behavior and symptom seems to
have been the foundation on which the concept of structure was built.

In the first phase of the theory (up to 1900), reality was considered
the factor which interfered with the drives, through the ego (con-
sciousness) in general and through its defenses in particular [96, 97].
But this view did not reach a conceptual status at the time and was
superseded (1900) by the conception of intrapsychic censorship exerted
by ego (self-preservative) instincts. A conception of psychological life as
a continuous clash of drive forces arose [98], and the abiding character
of the interfering factors was lost sight of. Not even the link established
between the concepts of censorship and the secondary process [98, 108,
117] conceptualized the abiding character of these two drive-controlling
factors. Instead, Freud again (1911-1917) became concerned with the
role of reality, and considered it to be the factor which interferes with
the drives and becomes a codeterminer of behavior. Yet he still assigned
the reality-testing functions to the secondary process and to the ego
drives [108, 121]. But the further study of censorship (particularly of
its repressive function) and of the secondary process shed additional light
on these interfering factors: they were now conceptualized as counter-
cathexes, which delay the discharge of drive cathexes, and by their
permanent deployment, prevent the return of the repressed [116, 117].
This formulation of a permanent deployment of countercathexes is the
beginning of the structural conception.

An explicit formulation of the structural conception became neces-
sary when it was realized that not only the drives but also most of these
invariant factors which interfere with drives are unconscious [126]. The

47 The structural givens in question are not the muscular apparatuses of motility,
nor the end organs of perception, etc., but rather their psychological regulations:
for instance, those psychological structures through which the control and trigger-
ing of the motor apparatus is effected.

The Structure of Psychoanalytic Theory 95

topographic conception, which relegated all crucial determiners (drives)
of behavior to the Unconscious and all epiphenomena and "apparent
determiners" (ego) to Consciousness, became systematically untenable,
and was replaced by the structural conception. The topographic division
into the Systems Unconscious, Preconscious, and Conscious yielded to
the structural conception of id, ego, and superego. The "ultimate drive
determiners" were conceptualized as id, the codetenniners (whether
conscious or not) as ego, and a specialized segment of the ego as
superego.48 Whereas the id was conceived of as the congeries of drives
(coexistent even if contradictory), the ego was defined as a cohesive
organization, whose function was to synthesize the demands of id,
superego, and reality [126]. The ego was conceived of as a structure
which codetermines (along with the drives) every behavior, and is re-
sponsible for the coordinated and organized character of all behavior, in-
cluding specific drive-discharge actions (e.g., sexual intercourse). But
the ego as a structure proved so complex that its exploration is even now
only just beginning. The recognition of the structure-building and
structural role of identifications [126] was followed by the recognition
of the role of the ego's defensive substructures [131]. In addition to these
two kinds of substructures, psychoanalytic ego psychology also came to
recognize orienting (perceptual), processing (conceptual), and executive
(motor) substructures, when it was realized that they are ready tools
(means) available to ego processes [157, 266].49

To begin with, psychoanalytic theory assumed that all psychologically
relevant structures arise in ontogeny. But at present some of these
structures are considered to be congenitally given. This shift has two
implications : first, that such constitutionally given apparatuses as motility,
perceptual system, memory system, thresholds50 are psychologically
relevant; second, that the ego does not derive from the id, but rather
both emerge from the common undifferentiated matrix of the first ex-
trauterine phase of ontogenesis [cf. Hartmann, Kris, and Loewenstein,
167].

While originally all structures were considered to be related to drive
and conflict, it is now assumed that the inborn ego apparatuses enter
conflicts as independent factors and that their function is not primarily
dependent on drives: thus they are termed ego apparatuses of primary

48 Of these three major structural concepts, in the following we shall discuss
only the ego. The structural treatment of the id and superego is still so inadequate
that the lengthy discussion it would require is beyond the scope of this presentation.

49 The fact that certain ego structures (e.g., defenses) have cognitive repre-
sentations does not contradict the distinction made here between defensive struc-
tures on the one hand and cognitive (means-) structures on the other.

50 Cf . note 47 above.

96 DAVID RAPAPORT

autonomy. [Cf. Hartmann, 157; see also, 266, 280.] This does not imply
that they have no relation to drives. They are part and parcel of the ap-
paratus which executes drive actions: they are the only means of action
the organism has. Nor does it imply that they are forever free of conflict :
under certain conditions they can and do become involved in conflict,
as does the motor apparatus in functional paralyses and the perceptual
apparatus in the tubular vision of hysterics. Thus even the apparatuses of
primary autonomy are only relatively autonomous from drive and con-
flict. But their autonomy does imply: first, that drives only trigger their
function and do not determine their course; second, that they can and
do function even when they do not serve the gratification of a specific
drive. Yet reservations must be made on both of these points: first,
while it is true that the role of drives in relation to these apparatuses is
primarily that of triggering their function, there is evidence to show that
the drives also have other effects on the apparatuses (e.g., the effects of
motivation on memory [258] ) ; second, the problem of the energy supply
of these apparatuses (when they are not triggered by drives) has so far
not been satisfactorily solved [161, 206, and 268, particularly part 7].
Woodworth's [329] conception of "habit as drive," Buehler's [42]
"pleasure in functioning/' Piaget's [254] "circular reaction," Allport's
[8] "functional autonomy" imply the same problem. Attempted solutions
either attribute drives (or partial drives) to apparatuses, or consider
apparatuses as sources of (neutral) ego energy, or assume that the energy
they use is neutralized drive energy at the disposal of the ego.

Psychoanalytic theory at first considered the structures which arise
in the course of ontogeny as conflict-born — i.e., defensive. Since defenses
are central to psychoanalytic therapy, they are the most extensively
studied structures, and this gives the impression that all structures are
conflict-born and all controls are defenses. Even though the role of
identifications in building ego structures was recognized early [126], a
tendency persists to consider this kind of structure-building, too, as
conflict-born. There is no theoretical clarity even now on this point
[see, however, Erikson, 66] : certain identifications definitely arise from
conflict (e.g., identification with the aggressor) ; others do not seem to.
But it is clear that means-structures born in, or used in, the course of
drive-gratification, or in the course of a defensive battle against drives,
or in the course of resolving a conflict can and often do undergo "change
of function"51 and become means of action and adaptation in the service
of the ego. These are termed structures of secondary autonomy [157,
162]. They, too, are only relatively autonomous in the same sense as are
the apparatuses of primary autonomy. They, too, are assumed to have

51 See Hartmann [157, 162]; for instance, rationalization is a defense mechanism
which tends to undergo a "change of function" and thus to become an important
means of adaptation, as a crucial ingredient of logical thought and rational action.

The Structure of Psychoanalytic Theory 97

neutral energies at their disposal or to use neutralized energies at the dis-
posal of the ego. The observations to which the concept of secondary
autonomy refers are akin to those which led to Allport's concept of
"functional autonomy" : they pertain to the relation of means-structures
to basic as well as higher-level motivations, attitudes, values, etc.

In conclusion:

1. The structural determiners of behavior were introduced as inter-
vening variables to account for the observation that motivations do not
determine behavior in a one-to-one fashion.

2. Structural determiners differ from motivational determiners in
that they are relatively permanent : their rate of change is relatively slow.

3. There are inborn structures and acquired structures: apparatuses
of primary and secondary autonomy.

4. Structure-building transforms motivations and thus gives rise to
new (more neutralized) motivations.

5. Structures built, and the motivations arising from them, may be-
come relatively autonomous determiners of behavior.

I. All Behavior Is Determined by Reality (the Adaptive Point of

View)

Reality in psychoanalytic theory designates the external source of
stimuli, including the subject's body, but excepting the somatic sources
of drives and affects [115, pp. 60-64]. In this theory reality (i.e., external
reality] is the antithesis of psychological reality™ [98, pp. 548-549].

This thesis of psychoanalytic psychology has undergone perhaps more
metamorphoses than any other, and its implications are far-reaching. It
implies, from the point of view of psychology, the question of the role
stimuli play in behavior; from that of biology, the question of the rela-
tionship between the organism and its environment; from that of philos-
ophy, the epistemological question (i.e., how man can know of, and act
in accord with, his environment when his thoughts and actions are de-
termined by the laws of his own nature) .

In psychoanalytic theory's first conception, reality was considered the
target of defense [94, ms. H; 96, 97]. More precisely, the defense was
directed against the memory of a real event, in order to prevent the re-
currence of the attendant feelings (affects) which, being socially pro-
hibited, were incompatible with self-respect™ Thus symptoms (i.e.,

52 The corresponding adjectives are real and psychological.

53 Self-respect and social prohibition did not attain conceptual status in psycho-
analysis proper. They cropped up in the early superego theory [114], and returned
in the Neo-Freudian schools. Only recently did ego psychology begin to work out
their place in the theory [Hartmann, 161; Bibring, 27; Erikson, 66; Jacobson, 190,
192],

98 DAVDD RAPAPORT

pathological behavior) resulting from defense (repression, etc.) were
considered to be ultimately determined by reality events. But the ques-
tion of reality's role in determining normal behavior was not yet raised,
although it was assumed that the affect of nontraumatic experiences is
"dissipated33 by being distributed over a wide associative network [cf.
35^ pp. 7_g; 'and 120] while the affect of traumatic experiences is

"dammed up."

The second conception of reality [98], which dominated psycho-
analytic theory from 1900 till 1923— with the exception of Freud's "Two
Principles" [108], which prepared the next conception— had two as-
pects: these were the drive object and the secondary process.

Drive was conceived of as an internal stimulus [115, 119] which,
unlike external stimuli, is continuous and inescapable through flight, a
stimulus for which the organism has no intensity-reducing barrier of the
sort which operates in regard to external stimuli. In turn external stimuli
were accorded little significance and psychological effectiveness, and no
behavior-determining role. Yet at the same time certain patterns of ex-
ternal stimuli, namely, drive objects, were conceived of as the precondi-
tion for drive action (drive discharge). Thus the effectiveness of drives,
as the ultimate determiners of behavior, remained in part dependent on
the availability of the drive object. Nor is this the only role reality plays
in this theory: configurations of reality which prohibit drive action were
considered to be represented intrapsychically by the censorship [98].
This is a drive-centered conception of reality: it comprises only the
conditions which make drive action possible or impossible. One feature
of this drive-object conception of reality has a broader significance.
While the instincts of animals on lower evolutionary levels appear to be
directly and more or less rigidly coordinated to specific external stim-
uli, the instincts of animals on higher evolutionary levels appear to be
less rigidly coordinated to such specific stimuli. This difference may be
characterized as a progressive internalization of the regulation of be-
havior.54 The psychoanalytic theory of drives assumes that the relation
of human drives to their drive objects is flexible, and that the regulation
of human behavior is to a large extent internalized [see 101]. Though
early psychoanalytic theory may at times have given the impression that
the organism is totally autonomous from its environment, it was never
so blind as to take this extreme stand. But it certainly does raise the
question of the organism's relative autonomy from its environment [cf . pp.
95-97, above; see also Gill and Brenman, 148; and 280], and does make
it clear that any explanation of behavior must come to grips with the
relative autonomy of behavior from both drives and external reality.

54 This internalization is considered coterminous with the establishment of the
ego; cf. Hartmann [157, 160].

The Structure of Psychoanalytic Theory 99

The other aspect of the conception of reality in this phase of psycho-
analytic theory was the secondary process. According to this theory, the
secondary thought processes tend to reflect reality "truly," i.e.', not
merely in terms of the drive whose object is to be reached, but in terms
of the "actual" relationships between objects which obtain in reality
[98, pp. 509, 533-536]. Likewise the secondary processes of action are
"adaptive to" reality [108, pp. 15ff.] and do not strive blindly toward
drive discharge. Delay of discharge, detour for the sake of a safe path,
"full" availability of memories and their use in the experimental action
of thought characterize the secondary process, which is thus not "selec-
tive" in the limiting sense that the primary process is but has a broad
access to reality over which it exercises selective judgments and choices.
This conception implied an "objective" reality, and secondary processes
which, unlike primary processes, do not "distort" but are "veridical,"
even though it was clearly recognized that the secondary process cannot
fully reduce these "distortions" because to do so it would have to elimi-
nate the affects which it needs as its orienting signals [98, p. 536]. This
conception remained incomplete, since it left the origin, nature, and
function of the secondary process unexplained [compare, however, Freud,
129, 130; also Ferenczi, 78, 79, 80, 81].

The third conception of reality appears in Freud's ego psychology, of
the 1923-1938 period, and was forecast in the "Two Principles" [108],
particularly by the concepts of reality principle and reality testing. In the
first conception, the defense was directed against reality and the memory
of real events. In the second conception, it was directed against the drive,
and reality had only a peripheral role. In the third conception, reality
and drive appear to gain a more or less equal status [131]. Now the
ultimate motive (determiner) of defense is real danger, and the drive
is defended against because if it were acted upon it would again lead
into a dangerous real situation. Thus defenses against drives come to
represent reality and, as constituents of ego and superego structure, they
become internalized regulations of behavior.

In the period we are now considering, the ego was still regarded as
a mainly defensive organization; nevertheless its origin in identifications55
and its most general definition56 point to its other functions and to its
intimate relation to reality. The identifications with the objects of social
reality imply that reality has not only a defensive-conflictful role, but
also an ego-structure-forming role. Moreover the ego, conceived as a
cohesive organization with a synthetic function of its own [131, pp. 25-
26; 245], gains a degree of independence from drives which permits a

55 "The ego is a precipitate of abandoned object cathexes" [126, p. 36], i.e., of
identifications with abandoned objects.

sa "^he eg0] is a coherent organization of mental processes" [126, p. 15].

100 DAVID RAPAPORT

relative objectivity in regard to reality. The relationship of the ego to
reality brings into sharp relief the central role of reality in this phase of
the theory: the ego's function is to reconcile the demands of the id,
superego, and reality [126]; the relation to reality is crucial to the ego
[137]; and the ego is organized around the System Perception-Con-
scious, i.e., around the means of contact with reality [126],

In this conception, reality shapes not only the ego, but even the
drives,57 which were previously conceived of as unchanging. Moreover,
in Anna Freud's [93, pp. 96, 109-110] conception, the defense against
reality itself again appears as a concept, much as it appeared in Freud's
first conception of reality.

The fourth conception of reality — Hartmann's — is a radical develop-
ment: the organism, as a product of evolution, is bom already adapted,
or potentially adapted, to reality. The ego apparatuses of primary
autonomy are instruments of and guarantees of man's "preparedness for
an average expectable environment." In animals of lower evolutionary
levels the instincts are the guarantees of reality adaptedness; man's drives
have lost much of this role, and thus inborn adaptedness is with him
more a potentiality than an actuality; processes of adaptation outweigh
inborn adaptedness. This potentiality for internalized regulation of be-
havior actualizes in the course of the development of the ego, which thus
becomes man's organ of adaptation.58

In this conception reality and adaptedness as well as adaptation to
it play a much more extensive role than in Freud's third theory [cf . Kris,
205; also Hartmann, 165]: here they are the matrix of all behavior.
Hartmann's concepts of relative autonomy., secondary autonomy, auto-
matization, and neutralization for the first time provide a framework for
understanding the development and the function of the secondary process
as one of man's major adaptative means. But Hartmann goes even
further and conceives of the reality to which man adapts as one created
by him and his predecessors. Yet even this conception seems to retain
an essential duality of psychological and external reality.

The fifth conception of reality, foreshadowed by both Freud's third
conception and Hartmann's, is the psychosocial one developed by Erikson
[61, 66]. Man is potentially preadapted, not only to one average expect-
able environment, but to a whole evolving series of such environments.
These environments to which man adapts are not "objective," but rather
social environments which meet his maturation and development half-
way: social modalities (e.g., the socially accepted forms of "getting")
foster, select, and harness his developing modes (e.g., the incorporative

5J See [139], but note this point already in [117].

58 Cf. Hartmann: Ego Psychology and the Problem of Adaptation [157]; see
also [160].

The Structure of Psychoanalytic Theory 101

oral mode) of behavior [62]. 59 This is the genetic counterpart of Hart-
mann's systematic formulation; it is thus far the only attempt to con-
ceptualize the phases of epigenesis [58, 61] through which preadapted-
ness becomes effective, and in which processes of adaptation inseparably
unite behavior epigenesis and environmental conditions [61, particularly
chap. 7]. The conceptions of an "objective" reality and of an unselective
and "veridical" secondary process disappear here and even the major
(time and space) coordinates of reality become "subjective" [63], in the
sense that they are shown to be relative both to organismic space and
time and to the particular society into whose reality man grows.60

J. All Behavior Is Socially Determined (the Psychosocial Point of

View)

To demonstrate that psychoanalysis considers real experience in
general and social experience in particular to be determiners of behavior
is to bang on an open door. For instance, organic psychiatry, which
centered on constitutional and hereditary factors, has always regarded
psychoanalysis as a pure "nurture psychology," but to do so, it had to
disregard the "nature" conception of drives in psychoanalytic theory. The
root of this and kindred misunderstandings seems to be that the theory
did not systematically clarify its stand on the dual relationship between the
organism and its environment. It is characteristic of organisms that they
are dependent on their environment but also relatively independent from
it. This balance between dependence and independence might be
designated as a relative autonomy (of the organism from its environ-
ment) in the same sense in which the analogous relationship of the ego
to the id is so designated [280].

The organic psychiatrists' view implies an absolute autonomy from
environmental influences. From their point of view, Freud completely
disregarded this autonomy, since he dealt mainly with the dependence

69 In this conception modes develop according to genetic, inborn laws, but the
social organization of the environment defines their place and form in the be-
havior repertory and their use in reality mastery and adaptation.

60 This conception does not deny the "objectivity" of the common, consensually
validated aspects of space and time, or the intellectual possibility of transcending
the subjective coordinates of reality in order to build universally valid sciences of
space, matter, etc. It does not invalidate the coordination of the organism with
the "objective" environment, which is guaranteed by the apparatuses of primary
and secondary autonomy (e.g., the perceptual and motor apparatuses), nor the
effectiveness of the "causal texture" of the environment (Brunswik, Heider) which
sets limits to all individual and social "subjectivity." For a detailed discussion of
these issues, which lead far into perception theory, see G. Klein's forthcoming
volume [198].

102 DAVID RAPAPORT

of behavior on experience, societal norms, etc. On the other hand, Freud's
stress on drives, as well as Ms regard for constitutional factors, made
his theory appear — even in the eyes of many of his followers — to imply
absolute autonomy from the environment. From this vantage point
society appeared as a factor interfering with nature, man was looked
upon as a born individualist, and therefore the therapeutic aim was
often regarded as the liberation of human nature from social bondage.

Ho\vever, Freud considered the sexual drives and their object choices
as anaclitic61 upon the drives of self-preservation and their object choices.
This is a statement of the growing organism's first social relationships
and implies the social determination of behavior. So does the conception
of the Oedipus complex: the budding individual's social environment
provides the objects of his libidinal and aggressive drives, and the
structures (identifications in ego and superego development) which
the relationships between the subject and these objects give rise to,
codetermine his behavior in general and not merely his pathology [126].
Though these social conceptions were not generalized into an explicit
psychoanalytic social psychology, the social determination of behavior
is clearly not alien to classic psychoanalytic theory.

Why then the reluctance of classic psychoanalysis to accept the
emphasis placed on the social determination of behavior by Adler,
Sullivan, Homey, and Kardiner? It seems that a struggle between dif-
ferent conceptions of the relative autonomy of behavior from environ-
mental reality62 lay behind this reluctance. To the classic analyst's
mind, the "dissident" schools, upon discovering the dependence of be-
havior on social reality, abandoned those concepts of the psychoanalytic
theory which encompassed the observations concerning the autonomy
of behavior from the environment: these were the drives and the other
constitutional (e.g., structural) givens. The net result seemed to be that
some dissidents came to regard adaptation as "adjustment" (particularly
as a therapeutic goal), to disregard the existence and nature of drives,
to stress the environmental demand, and thus deliberately or unwittingly
to reinforce censorship and superego. These therapists were said to have
come to take "society's side" against the patient — although their task,
as originally conceived, was to take neither or both. In other dissidents
the result seemed to be very different: society was blamed for man's
troubles and was to be so changed as to cause no more trouble to man.
This was said to be a stand on the side of the individual against society.
Thus, do-goodism, social rebelliousness, Philistine demands for con-

61 Anaclitic: leaning upon. The implication is that the first objects of the
sexual drive are the people who take care of the infant and guarantee his survival,
i.e., who are the objects of his self-preservative drives [101],

62 Gf.5 for this section, Fenichel's discussion of Fromm [76].

The Structure of Psychoanalytic Theory 103

formity, and attitudes so approving of individual liberty as to shade
into license were considered to be characteristic of the dissidents. Any
of these attitudes may well be a therapist's private convictions, and
as such cannot but enter therapy in some form, but none of them has a
place in psychological theory. It is not our task to establish whether
these are fair assessments of the dissident schools.63 It is sufficient to
point out that these imputed or actual attitudes are indices of a struggle
centering around the ego's relative autonomy from social reality. The
dissidents' contribution toward the formulation of the psychosocial point
of view must not be underestimated.

Anna Freud's stress on work with the parents of her child patients,
and her work with groups of children in the course of the war, is a
recognition of the social determination of behavior.

Hartmann gave the first theoretical formulation of the role of social
reality. His point of departure was Freud's conception of the central
role in human development of the infant's prolonged helplessness and
dependence on caretaking adults. Hartmann's major theoretical advance
is embodied in his concept of "social compliance," which is coined on
Freud's concept of "somatic compliance." The referents of this concept
are the observations concerning those institutions of society which meet,
foster, and mold the developing individual's inborn and acquired
adaptive means. Education as a social institution is an instance of "social
compliance" [157],

Erikson's epigenetic psychosocial conception parallels and fills in
Hartmann's systematic and programmatic formulations. Erikson's con-
ception of society is detailed : it is the geography and the means of sur-
vival it provides; it is the economy and other social institutions; it is the
ideology, including tradition [61, chaps. 3, 4; also 60]. It involves: (1)
the epigenetic conception of ego development consisting of a sequence of
developmental phases, each characterized by a phase-specific crisis which
is universal, while its solution varies from society to society and is in-
dividually unique [61, chap. 2; 62, 66]; (2) the caretaking people
(parents, teachers, etc.) and their practices representing the society's in-
stitutions and traditions which were developed to meet each phase-specific
crisis of the developing individual's life cycle; (3 ) the phase-specific needs
of the growing individual eliciting dovetailing needs in the caretaking
people, which correspond to the respective phases of their life cycles ; the
society's institutions and traditions making their contribution to the
solution of the growing individual's phase-specific crisis by means of the
so-elicited needs of the caretaking people; (4) the resulting behavior

63 FenicliePs assessment [76] of Kardiner and Fromm seems as adequate as the
conceptual equipment of the time permitted: Hartmann's and Erikson's concepts
had not yet entered the argument.

104 DAVID RAPAPORT

forms having, by and large, an accepted place in the society and guaran-
teeing the individual's viability in it.

In Erikson's conception neither does the individual adapt to society
nor does society mold him into its pattern; rather, society and individual
form, a unity within which a mutual regulation takes place. The social
institutions are preconditions of individual development, and the develop-
ing individual's behavior in turn elicits that help which society gives
through its adult members directed by its institutions and traditions.
Society is not merely a prohibitor or provider; it is the necessary matrix
of the development of all behavior. Indeed, the development and
maintenance of the ego, of the superego, and perhaps of all structures
are dependent on the social matrix : behavior is determined by it and is
possible only within it.

BetteUieim's [21, 22, 23] and Redl's [282, 283] studies confirmed
this conception and extended it. Gill and Rapaport [149] concluded,
from the observations and theories here discussed, that the meta-
psychological triad of the dynamic, structural, and economic points of
view must be extended by the addition of an adaptive point of view.

K. Discussion

This sketch of the basic propositions of the general psychoanalytic
theory was presented to make a discussion of its variables possible. It
centered around the three classic metapsychological points of view
(dynamic, topographic, and economic) [cf. 117, p. 114], but it also in-
cluded the structural point of view (which elaborated and replaced the
topographic one) as well as the genetic and the adaptive points of view,
which (being of the same order of significance in the theory as the classic
triad) seem necessary to complete the system of psychoanalytic metapsy-
chology [see Gill and Rapaport, 149]. The inclusion of the psycho-
social point of view (like that of the topographic one) is a mark of
systematic weakness, since it is merely a specific aspect of the adaptive
point of view. It is as yet difficult, if not impossible, to present the
theory divorced from its history. The topographic point of view, though
it is satisfactorily replaced by the structural one, appears here because it
is difficult to present the latter so that the role of unconscious deter-
mination will emerge as clearly as it does from the by now historical
topographic point of view. Likewise, the psychosocial point of view is
discussed separately because it is as yet difficult to present the adaptive
point of view so that its psychosocial implications emerge clearly. Both
Hartmann's and Erikson's theories are too new, their implications too
little understood, and their relationship to each other too little ex-
plored [see Rapaport, 277, 278] to permit a statement disregarding all
but systematic considerations.

The Structure of Psychoanalytic Theory 105

Thus it is likely that seven of these ten points of view which we
have discussed here will, in future systematic treatments, be condensed
into the five (dynamic, economic, structural, genetic, and adaptive)
metapsychological points of view. Indeed, they may eventually be
formulated as the axioms of the system. The remaining three points of
view (empirical, Gestalt, and organismic) seem to be of a different
character, and lumping them together with the metapsychological points
of view is another indication that the systematization here attempted is
premature. By and large, they deal with the theory's observables and
with their organization as units. It is possible that these three points of
view will appear, in future systematic statements, as definitions of observ-
ables. If so, then why were they not segregated here from the others?
One reason is that the others are not yet formulated as axioms and they,
too, imply definitions. Then why not explicate all the definitions and
segregate them from the points of view? At this stage of our knowledge,
even if such explication and segregation were possible, the present analysis
would be an indispensable preparatory step.

Tolman's [316], MacGorquodale and Meehl's [227], and others'
conceptions of independent, intervening, and dependent variables can-
not be discussed here in detail. It should suffice to say that their views
of these variables do not seem to be "methodologically pure," but
rather loaded with their respective systematic biases.63a Only that con-
ception of variables which Koch calls the "mathematical" seems relevant
to psychoanalytic theory. It is in the sense of such a mathematical con-
ception that we will speak here of variables.

I believe that the following conclusions may be derived from the
sketch of the theory's "points of view" :

1. The psychoanalytic concept of over determination implies that one
or several determiners of a given behavior, which appear to explain it,
do not necessarily give its full causal explanation. This is not per se alien
to other sciences, though a principle of over determination did not be-
come necessary in any of them. Psychoanalysis' need for this principle
seems to be due partly to the multiplicity of the determiners of human
behavior, and partly to the theory's characteristic lack of criteria for the
independence and sufficiency of causes. The determiners of behavior in
this theory are so defined that they apply to all behavior and thus their
empirical referents must be present in any and all behavior. Since there
is usually no single determiner which constantly assumes the dominant
role in a given behavior, other determiners can hardly be neglected
while a dominant determiner is explored. When favorable conditions
make one determiner dominant, the investigator is tempted to conclude

63a Frenkel-Brunswik [92, pp. 307ff.] gives a cogent discussion of some of these
biases. '

106 DAVID RAPAPORT

that he has confirmed a predicted functional relationship — as he indeed
has. Regrettably, the attempt to repeat the observation or experiment
in question often fails, because in the replication either the same be-
havior appears even though a different determiner has become dominant,
or a different behavior appears even though the same determiner has
remained dominant.64 Lewin's Gesetz und Experiment in der Psychologic
[215] is relevant here: it argues that the criterion of validity for psy-
chological experiments [cf. Gill and Brenman, 34, 147; and Benjamin,
1 1] is not repeatability, but predictable systematic variation.

The implications of the concept of overdetermination for the choice
of independent variables are: (a] Any motivation high in the hierarchy
of psychological organization, if chosen as the independent variable of
an experiment or observation, may prove to be dependent on variables
closer to the base of the hierarchy. In this case, either the dependent
variable will be treated as an implicit function of these more basic varia-
bles (motivations, structures, etc.), or the latter will be considered as inter-
vening variables interposed between the independent and the dependent
variable, (b) If a basic motivation is chosen as the independent variable
of an experiment, then variables higher in the hierarchy will be
interposed as intervening variables between the experiment's independent
and dependent variable. For instance, in Klein's [197] experiment,
cognitive attitudes are the intervening, thirst the independent, and
cognitive behaviors the dependent variables. Though in this theory basic
drives are systematically distinguished65 independent variables, as em-
pirical independent variables they do not seem to differ significantly from
other motivational variables.

2. The psychoanalytic conception of autonomy puts a further limita-
tion on the distinguished independent variable character of basic drives
by pointing to other equally distinguished ones. The concept of autonomy
implies that structures of primary (and secondary) autonomy may re-
tain (or attain) a relative independence from drives. The function of
autonomous structures, even when triggered by drives, may remain in-
dependent from them. Derivative motivations (as a rule, related to

64 For instance, in a Zeigarnik-like experiment, on the one hand an interrupted
task may be remembered either because of the undischarged tension system
(Lewin's explanation) or because the task had a specific "historical" or "motiva-
tional" significance for the subject to begin with. In this instance different domi-
nant determiners have identical effects. On the other hand, an interrupted task
may be forgotten (in spite of the undischarged tension system) when the inter-
ruption is experienced as a failure. In this instance the undischarged tension re-
mains the dominant determiner, but its behavioral effect is different.

65 The term "distinguished" is used here to convey that though the theory's
development placed restrictions on the initial conception of drives as "ultimate
causal determiners," drives still retain a special position in the system of the
theory (see pp. 89-91, above).

The Structure of Psychoanalytic Theory 107

structures) may also attain such independence. The implications of
this concept of autonomy, for the choice of independent variables, are
contrary to those of overdetermination : since autonomy increases with
the distance from the basic drives, variables (structures or motivations)
high in the hierarchy also appear as systematically distinguished in-
dependent variables. As independent variables of an experiment, they
have the advantage that the dependent variable need not be an implicit
function of more basic ones, and that (in the ideal case) no intervening
variables are interposed between them and the dependent variable.

The conflicting implications of the "overdetermination" and the
"autonomy" concepts limit the advantages of the latter also, since not
all structures and derivative motivations retain or attain autonomy; and
moreover, autonomy is not an all-or-nothing affair, but rather a matter
of degree, and thus we always deal with relative autonomy, the degree
of which must be empirically determined.

3. Basic motivations and structures, as well as motivations of a high
degree of autonomy, are systematically distinguished variables. Whether
they should be considered to be systematic independent variables (in
Koch's sense) is not clear, since they may also appear in the role of inter-
vening and dependent variables. To illustrate this, let us survey the main
classes of the theory's variables: motivations and structures (of any
hierarchic level and degree of autonomy), behaviors (including thought
and affect as well as observable action), and external reality.

External reality. In the reflex-arc model external reality (stimulus)
appears as the independent variable. The model assumes that in this
case unconscious impulses and ideas always enter as intervening variables,
and that the dependent variable is motor action and/or conscious
thought and/or affect. However, the autonomy concept implies that
the functional relationship between stimulus and behavior may be of any
degree of relative autonomy; i.e., the extent to which unconscious im-
pulses and ideas intervene may vary. Thus S-R psychology appears
here as a limiting case of a high degree of autonomy (automatization).
[See Hartmann, 157, pp. 26, 86ff.]

External reality as an intervening variable is one of the implications
of the adaptive point of view. When either structure or motivation is
chosen as the independent, and behavior as the dependent variable,
external reality appears as the intervening variable, and corresponds to
the adaptive aspect of the behavior in question.66 The concept of relative
autonomy from the environment, however, implies that some of the
motivation vs. behavior and structure vs. behavior relationships (like
those in impulsive actions, fugues, and characterologically typical be-

66 For instance, in studying the effect of hunger on feeding behavior, the absence
of, or presence and demeanor of, an observer will enter as an intervening variable.

108 DAVID RAPAPORT

haviors respectively) will be, within limits, invariant in regard to
changes in external reality.

How external reality can be a dependent variable is less obvious. If
external reality were conceived "geographically" [see Koffka, 204, pp.
27ff.]5 then it would be meaningless (or a subject matter for physics,
chemistry, etc. ) to treat it as a dependent variable. If, however, we go be-
yond its inherent "causal texture" (Brunswik, Heider) and conceive of it
psychologically, then it can become a dependent variable. A person in
my life space is an external reality, yet this external reality is a variable
dependent on my "feelings" toward this person. In Bruner and Good-
man's experiment, the sizes of coins appear as dependent variables,
though in that instance it is difficult to separate perceptual behavior and
external reality as dependent variables.

Motivations. In deprivation experiments, as well as in observations
made in therapy (e.g., on transference phenomena), motivations appear
as independent variables ;&7 and their hierarchic position (implying con-
siderations of overdetermination and autonomy) defines the degree of
their actual independence. In these instances defenses and other struc-
tures usually enter as intervening variables, and behavior is the most
common dependent variable, though in investigations concerning proc-
esses of structure building and structural change, structure will be the
dependent variable.

In the reflex-arc model, unconscious motivations appear as inter-
vening variables, and external reality plays the role of the independent,
and behavior that of the dependent variable. Moreover, wherever a con-
trolled (not impulsive) motivation, or a structure close to the base of the
hierarchy of the mental organization, is taken as the independent
variable, higher-level motivations appear as intervening variables, pro-
vided that no automatized relationships obtain between the independent
and the dependent variables.

Motivations as dependent variables are encountered when motiva-
tions close to the base of the hierarchy are chosen as independent vari-
ables and defensive structures appear as intervening variables; or when
external reality, in the form of deprivation, is the independent variable.
Clinically the presence of defensive structures is — as a rule — inferred
from the appearance of derivative motivations,68 which are in this case
dependent variables. But certain motivations may also appear as de-
pendent variables where external realities (other than deprivation)
or psychological structures are the independent variables.

6T Except where the degree of deprivation is the independent, and the drive the
dependent variable.

68 For instance, in studying a coprophilic impulse, the presence of reaction
formation may be inferred from the appearance of a motivation for excessive
cleanliness.

The Structure of Psychoanalytic Theory 109

Structures. Structures appear as independent variables wherever in-
dividual differences in behavior, under (relatively) constant motivation
and stimulation, are studied: for instance, in the comparative study of
symptoms in various neuroses, and in the studies of individual differences
in perception [Klein, Holzman, Gardner, Schlesinger, 144, 179, 180,
198, 304].

Structures as intervening variables are commonplace in clinical ob-
servation. They account for the lack of a one-to-one relationship between
motivations and behavior. Defensive structures countermand motivations
and replace them by derivative motivations (as, for instance, in reaction-
formation). Controlling structures direct and channel motivations, as
in delay and detour-behavior and in the choice of substitute goals. In
Klein's [197] thirst experiment, the thirst motivation was the inde-
pendent, perceptual behavior the dependent variable, and structures
(the subjects5 "cognitive attitudes") appeared as intervening variables.

It is less easy to conceive of structures as dependent variables,
though they do appear as such in processes of structural change, includ-
ing those of learning. In so far as psychoanalysis as therapy achieves its
goal of changing existing structures, in at least some of the observations
made in therapy, structures appear as dependent variables. Piaget's
[254] schemata of the primary, secondary, and tertiary circular reactions
are structures, and in his developmental studies which trace their growth
and fate, structures are dependent variables.

Behavior. The role of behavior as a dependent variable needs no
discussion. But it might be worth noting again that here behavior is
broadly defined to include -conscious and unconscious thought, affect,
and action, which can and do substitute for one another, so that behavior
is a complex dependent variable.

The role of behavior as an intervening variable is more difficult to
conceive of, though it is commonly enough encountered. When, for ex-
ample, a motivation is taken as the independent variable and the ob-
servable action facet of behavior is taken as the dependent variable,
the thought and affect facets of behavior, as a rule, interpose them-
selves as intervening variables. This seems to be one of Hebb's [169]
points in his criticism of S-R theories. Naturally, in impulsive actions
and where the relation between motivation and overt action is auto-
matized, such intervening variables are likely to be absent.

The conception of behavior as an independent variable is perhaps
the least obvious of all. Yet, for example, under conditions of a high
degree of autonomy, one facet of behavior may be taken as an in-
dependent and any other facet of it as the dependent variable. For in-
stance, in Werner's experiments [152, 208] — in which the subject
presses against motor restraint, with the consequence that the number
of his movement responses in the Rorschach test increases — the motor

110 DAVID RAPAPORT

facet of behavior is the independent, while its cognitive facet is the
dependent, variable.

4. It seems that the variables of all these major classes can be treated
as empirical independent, intervening, and dependent variables; but
they differ greatly in regard to manipulability, which is considered by
some to be the criterion for the selection of independent variables.
Genetic and structural variables, for instance, are not amenable to direct
manipulation. Besides such intrinsic difficulties, manipulation of the
internal and external environments as well as of action is also limited
by due regard for the subjects' privacy and by the fact that social
manipulation beyond a narrow range is likely to endanger the in-
dividual's rights. But manipulability is not an indispensable criterion;
it may be replaced by observation (as in astronomy), or by seeking out
"nature's experiments" (as in evolution theory) .

5. We may conclude that psychoanalytic theory requires the ex-
ploration of all the possible functional relationships among its variables.
One wonders whether or not there is an intimate connection between the
rigid decisions of various schools of psychology on systematic variables
(e.g., those of S-R and Gestalt psychology) and the limited range of
observables acceptable to each of them. Any limitation on the choice
of variables seems to result in a limited range of observables and ob-
servational methods, and it is the dearth of methods which is probably
the major obstacle to bridging the gap between psychoanalysis and
academic psychology [cf. Shakow and Rapaport, 309], and between the
various schools of psychology.

III. THE INITIAL EVIDENTIAL GROUNDS
FOR THE ASSUMPTIONS OF THE SYSTEM
AND THEIR STRATEGIC CHARACTER

A. Initial Evidential Grounds

We will discuss here only the evidential grounds for the early as-
sumptions of the system;69 to trace those of its present assumptions would
be a historical job far exceeding the scope of this essay. Thus, the
propositions to be discussed in this section are not always identical with
those of the present theory.

The basic assumption of psychoanalytic theory was and is thorough-
going psychological determinism. Its other initial assumptions are im-
plicit in the thesis of psychoanalytic metapsychology: a full description
of any psychological phenomenon must include its dynamic, topographic,

69 The reference here, if not otherwise indicated, is to Breuer and Freud [35].

The Structure of Psychoanalytic Theory 111

and economic descriptions.70 What are the assumptions implied in these
three points of view?

The topographic point of view distinguishes between the Systems
Unconscious, Preconscious, and Conscious, and thus implies the as-
sumption of unconscious psychological processes,, which is, except for
psychological determinism, the earliest and most general assumption of
psychoanalysis. On this assumption are built the concepts of unconscious
motivation (invoking dynamic assumptions) and primary process (in-
volving economic assumptions).

The dynamic point of view implies the assumption of psychological
forces and their conflicts in general and of the drive nature of these
forces in particular [98]. On these assumptions are built the concepts of
libidinal drives and censorship (ego drives or self-preservative drives —
which are now dated), as well as the conception of the central role of
libidinal drives [101].

The economic point of view implies the existence of psychological
energies in general and their drive origin in particular. These assump-
tions underlie the concept of cathexis (quantity of energy). Corollaries
of these assumptions are principles analogous to the physical principles
of conservation of energy, entropy, and least action. The conservation
principle: cathexis is never lost and thus is traceable in the expenditures
and transformations of cathexes involved in psychological forces [117,
p. 114]. The entropy principle (the much misconstrued pleasure
principle) : drive energy tends toward discharge (i.e., diminution of
tension) [35, p. 143; and 98, pp. 508-509, 533-535]. The principle of
least action: processes involving cathexes other than those of basic
drives operate so as to expend the least amount of cathexis [98, pp. 533-
534]. The main concepts built on these principles are wish-fulfillment
vs. reality-testing which direct and the primary vs. secondary process
mechanisms which subserve the transfer and transformation of cathexes
[98, pp. 530-531, 535-536].

The initial evidence for these three sets of assumptions and their
corollaries cannot at this time be sharply separated from the evidence
for the validity of the theories built upon them. A sharp separation would
require prior decisions as to which assumptions are to be treated as
axioms and which are to be empirically derived from a combination
of axioms, definitions, and observations.

1. The assumption of psychological determinism. The initial evi-
dential ground for this assumption was the observation that apparently
meaningless hysterical symptoms, previously attributed to a somatic

70 This is the earliest explicit formulation of metapsychology; see Freud [117,
p. 114].

112 DAVID RAPAPORT

etiology,71 disappeared when the patient, in hypnosis, related them to past
experiences, thoughts, feelings, or fantasies, and thus endowed them with
meaning and psychological "cause."72 This success at tying apparently
arbitrary pathological behavior into a causal psychological network
served as the empirical point of departure for the venture into the
broader realm of dreams [98], parapraxes [99], etc. The success in
"interpreting" these resulted in the further and apparently limitless
generalization of this assumption, on which all the other initial assump-
tions of the theory rest. Clearly, the empirical evidence alone, without
the background factors discussed in the introduction of this essay, might
not have given rise to the assumption of psychological determinism.

2. The assumption of unconscious psychological processes. The ob-
servation that in hypnosis and in the course of free associating patients
become aware of past experiences, or of relations between them, or of
relations between past and present experiences, led to the assumption
of the "nonconscious" survival of such experiences and the "non-
conscious" existence of such relationships [35, 95]. But only the discovery
that such nonconscious experiences and relationships are subject to rules
(e.g., the pleasure principle and the mechanisms of the primary process)
different from those of our conscious behavior and thinking made the
above-mentioned memory phenomena (already observed by Charcot, as
well as Bernheim) [see Breuer and Freud, 35, chap. 1] into evidence for
the assumption of unconscious psychological processes [98]. The essence
of this assumption is that it conceptualizes these observations in psy-
chological terms, though the processes inferred from them are subject to
rules different from those of the familiar, conscious psychological proc-
esses. In other words, it refuses to treat the nonconscious as somatic and
the nonlogical as nonpsychological. It rejects both consciousness and
logical relations as necessary criteria of psychological processes, and thus
arrives at the concept of unconscious psychological processes abiding by
rules other than those of conscious processes. This assumption gained
powerful corroborative evidence from the study of dreams [98, p. 540].

3. The assumption of unconscious psychological forces and conflicts.
The evidence for unconscious psychological processes did not, in the
beginning, necessitate the assumption of unconscious psychological
forces and conflicts. Breuer's hypnoid assumption and Freud's trauma

nEven the psychologically minded French school, Charcot, Janet, etc. [see
309], subscribed to this.

72 For the detailed reports of these observations, see [35]. The theoretical section
of that volume contains a fragmentary and simplified version of the neuropsy-
chological theory Freud developed in the Project [94, appendix] to account for
these observations. These two theoretical statements are the predecessors of Freud's
theory contained in the seventh chapter of The Interpretation of Dreams [98].

The Structure of Psychoanalytic Theory 113

and retention assumption [35] seemed to account for the unconscious
character of these processes. The evidential ground for the assumption
of unconscious psychological forces was Freud's discovery that much
of what his patients reported to him was not, as he initially believed,
unconscious memories of actual experiences, but rather unconscious
fantasies [94, p. 215, letter no. 69, 1897]. The assumption of unconscious
forces was to account for the agent which creates these fantasies and
brings them to the patient's consciousness in hypnosis and in free as-
sociations, as well as for the agent which, before and in the course of
therapy, prevents them from becoming conscious. Led by the libidinal
content of these fantasies, Freud assumed that the unconscious force
which creates them, and makes them conscious in the course of therapy,
is the sexual drive. In turn, he conceived of the forces which clash with
the sexual drives, divert them into symptoms, and block the path to
consciousness of the fantasies which they gave rise to, as those of the
censorship : the ego drives.

Thus the initial evidential ground for this assumption comprised
observations pertaining to unconscious fantasies, to their becoming con-
scious in therapy, to the resistance against their becoming conscious, and
to the relation between these fantasies and the symptoms.

4. The assumption of psychological energies and their drive origin.
The observation that recall of traumatic experiences, when accompanied
by affect, results at times in the disappearance of symptoms and anxiety,
and at other times in their replacement by other symptoms and anxiety
equivalents, suggested that a displaceable and transformable quantity
was involved in the psychological processes underlying symptom forma-
tion. Before he developed the concept of unconscious forces, Freud as-
sumed that this quantity was the affect, which when not expressed (i.e.,
"dammed up53) was either transformed into anxiety or displaced into a
somatic organ (conversion) or a thought process (e.g., obsession). After
he developed the concept of drives, this quantity was conceived of as
drive energy (cathexis).

The force concept alone could not account for the observation that
blocking a drive action results in behavior different in direction and
form from that expected of the drive; this observation became the
evidential ground for the assumption of psychological energies and of a
conservation principle pertaining to them. These psychological energies,
analogous to those of physics, being nondirectional (scalar) could,
through their displacements and transformations, account for the "work"
performed by the psychological force in forms unlike, and at points not
coinciding with, that expected of them. This assumption when combined
with that of the instinctual origin of the unconscious psychological forces
led to the assumption of the drive origin of psychological energies.

114 DAVID RAPAPORT

The evidence for the assumption of an entropy principle and a
principle of least action lay in the observation of the difference between
those two kinds of behavior characteristics which were conceptualized as
manifestations of the primary and secondary processes. The prevalence
of the first kind of characteristic makes a behavior peremptory and over-
valent, tolerating neither delay nor detour, as though it were striving
for immediate discharge of a great quantity of excitation. These char-
acteristics of obsessional and delusional ideas, compulsive rituals, hys-
terical tantrums, etc., served as the evidential ground for the assumption
of an entropy (pleasure-pain) principle. The prevalence of the second
kind of characteristic makes a behavior amenable to interruption, delay,
and detour, as though it involved potentials without significant intensities.
These characteristics of goal-directed action, and of ordered logical
thought, were the initial evidence for the assumption of a principle
of least action.

It may be objected that this discussion has not dealt with the as-
sumption of the ultimate determination of all behavior by unconscious
drives, which, with its emphasis on ultimate and all, is indeed one of the
assumptions of early psychoanalytic theory. We bypass it here because it
is actually a combination of the assumptions we have discussed.

B. Strategic Choice of Initial Evidential Grounds

The question why the observations which served as the initial evi-
dential ground for the assumptions of the system were considered
strategic is in a sense irrelevant to the theory of psychoanalysis. The initial
situation was not that the phenomena of pathology were considered
strategic: they were the material which posed the problems to be coped
with. The theory grew up on the soil of the neuroses, their pathology
and their therapy. It was from there that it branched out into a relentless
and ever more diversified endeavor to show that its system of conceptual
relationships, though it was designed to explain pathological (apparently
arbitrary and psychologically meaningless) phenomena, can also give an
adequate causal account of the obviously meaningful phenomena of
normal psychological life.

Pathology was (as Virchow recognized in biology) strategic for the
study of normal processes.73 It showed that the so-called normal state of

73 Pathology had still another role in the development of psychoanalysis: ex-
cept for love and mortal fear, only actual suffering and the hope of relief could
have prompted a man to permit another that relatively unlimited access to his
privacy which opened the door for psychoanalysis to the exploration of its initial
evidential ground. That this opening is at the same time an obstacle to the inde-
pendent verification of psychoanalytic theories is as -natural as it appears para-
doxical at first sight.

The Structure of Psychoanalytic Theory 115

affairs which we take for granted is only one of many possibilities. Thus
it opened the road to causal analysis, by means of which psychoanalysis
consistently and successfully shattered the barriers between the normal
and the pathological, the infantile and the adult, the recondite and the
obvious, the exceptional and the commonplace [98, pp. 538-540]. It is
not a historical accident that Freud's theory grew out of the study of
pathology.

Pathology and its therapy were strategic for the discovery of the
commonalities of normality and pathology, but they proved less strategic
for the discovery of the differences between them. Only slowly, with the
development of psychoanalytic ego psychology, has psychoanalysis begun
to rediscover the differences between the pathological and the normal,
the infantile and the adult, the maladaptive and the adaptive. So far,
the concepts of structure, autonomy, adaptation, and reality are the main
tools the theory uses in its endeavor to discover these differences. These
are the very concepts which distinguish psychoanalysis from the genetic
reductionist theories which see no cleavage between the normal and the
pathological, the adult and the infantile, as well as from G. AUport's
[8] and kindred theories which see a sharp cleavage between them.

We cannot leave this discussion without dwelling, at least briefly,
on the methods by which the initial evidence was obtained.

Nowadays methodology is in vogue, and all considerations of method
and technique are dignified by that name. Yet one essential methodo-
logical task — the study of the relationship between a theory and the
method of observation by which the data it explains are obtained —
is rarely pursued. The question is: to what extent does a theory, based
on data obtained by a given method, reflect the nature of the data itself,
and to what extent does it reflect the method of data-gathering and its
limitations? The man who shouts into an empty room is likely to hear his
own echo; likewise the investigator may get back little more than what
he has already built into his method. For instance, we need to know to
what extent the "laws of learning" are laws of human nature, and to
what extent they are artifacts of the method used by associationists and
conditioners to "establish" them. Likewise to what extent does psy-
choanalytic theory reflect human nature, and to what extent does it
reflect Freud's methods for studying human nature?

Methodological study is likely to reveal that some psychoanalytic
methods (for instance, the therapist-patient two-group) had a defining
influence on psychoanalytic theory [see Rapaport, 260]. Although we can-
not pursue this problem further, we want to suggest that methodological
analysis may well lead to a distinction between a general psychoanalytic
theory which is little dependent on these methods, and a specific psycho-
analytic theory which is greatly dependent on them. In contrast to the

116 DAVID RAPAPORT

specific theory, the general theory should be readily testable by methods
other than those by which the initial evidence for it was obtained. In this
essay we have centered on those aspects of the theory which are not
obviously dependent on these methods, and have tried to avoid the
concepts which obviously are tied to them, like transference, interpreta-
tion, etc.74

C. The Relation of the Observations to the Theory

In his outline, Dr. Koch asks us to select the chief empirical in-
dependent and dependent variables of the theory and to demonstrate
their linkage to its systematic independent and dependent variables. In
Section II. K., we discussed the difficulties in the way of such an at-
tempt. To minimize repetition, we will illustrate75 the relation of an
empirical observation to the variables of the theory.

Let us take the words of a man who utters the sentence, "Now things
are becoming queer,53 and let us provide the context from which psy-
choanalytic theory will adduce its explanation of this verbal behavior:

This behavior occurred in the course of a discussion in a group. — The
other members responded to it with consternation. — The man was be-
wildered by this response. — Later he found out that he had said "queer"
and not "clear" as he had intended to do, and as he thought he had
done. — He was embarrassed by this discovery. — The discussion concerned
a mismanagement of the group's affairs. — The subject's utterance fol-
lowed an explanation by the chairman of the group. — The chairman at-
tributed the mismanagement to a misunderstanding by the treasurer of
an instruction given by him, and not to any malicious intent. — The chair-
man commanded the unquestioned respect of the group and also wielded
considerable power otherwise.

In terms of common-sense psychology, we are dealing here with a
slip of the tongue.

In descriptive terms: the subject's conscious intention was to agree
with the chairman's explanation. He did not carry out this intention, but

74 An example to highlight the relationship of method and theory: it appears
that H. S. Sullivan [314], taking as his point of departure the psychoanalytic meth-
ods of the two-group and participant observation, arrived at a theory of per-
sonality which dissolves the concept of the individual and conceives of the person
as one of the quasi-stable foci in the network of interpersonal relationships. In
Sullivan's theory then, the method of investigation and the transference concept
based on it came to play a dominant role, with the consequence that the theory
overrides a crucial characteristic of the nature of the subject matter, namely, the
individuality of the person. Individuality to Sullivan appears as a noxious anti-
scientific myth, which he reduces to the personification function of the self-system.

's This example simplifies an actual situation by eliminating obscure points, to
avoid lengthy explanations of peripheral matters.

The Structure of Psychoanalytic Theory 117

instead expressed himself in a way that caused consternation; he was
unaware both of not having carried out his intention, and of his con-
sternation-arousing utterance. When he was told what he had said, he
became embarrassed.

In terms of a data language: the independent variable (conscious
intention) determined a value of the dependent variable (verbal
utterance of agreement and conscious awareness of it). However, an
intervening variable determined another value of the action component
of the dependent variable (dissenting, consternation-arousing verbal
utterance). The intervening variable left the conscious-awareness com-
ponent of the dependent variable unaltered. A second intervening vari-
able (external reality: information) altered the conscious-awareness com-
ponent of the dependent variable and determined its affective aspect
(embarrassment) .

This formulation is not "neutral": it implies that the independent
variable is a "motivation" (intention). Indeed, even the descriptive
terms imply this. Before Freud, at least the common-sense term "slip of
the tongue" was neutral, but it is not neutral now. Let us attempt a
crude associationist formulation, to show that data languages are in-
separable from construct languages and thus cannot be neutral: the
chairman's explanation was associated in the subject to an approving
verbal statement; the subject's actual response, however, was linked to
the chairman's explanation by stronger associative bonds; the clash of the
two associative complexes resulted in a compromise in which one of the
complexes determined the awareness, while the other determined the
verbal response of the subject.

In terms of psychoanalytic construct language: the subject's con-
scious intention is referred to a socially adaptive ego interest. The failure
to carry out the intention is referred to an id motivation. The unaware-
ness of the failure is referred to an unconscious ego motivation conflicting
with this id motivation. The unawareness of the actual verbal expression
used is referred both to the unconscious (id) motivation which was ex-
pressed, and to the unconscious ego controls (defenses) which, though
they failed to prevent the use of the ego's executive apparatus by the id
motivation, succeeded in preventing its access to consciousness (com-
promise). The acute embarrassment is referred to the affect manifesta-
tion of the clash between the unconscious motivations and the restored
ego control.

Let us take a closer look at the concepts involved. The unawareness
is obviously the referent of the descriptive concept unconscious. It is
likewise obvious that the intent to say "clear" is a conscious motive.
But it is an inference that this motive is a force and it is a further in-
ference that saying "queer" indicates the presence of another, un-

118 DAVID RAPAPORT

conscious, force. It is a still further inference that a third force is also
involved which prevents conscious awareness both of the failure of the
conscious intention and the success of the unconscious intention. It is yet
a further inference that the latter two forces conflicted and reached a
compromise, indicated both by the "clang" similarity of the words
"clear35 and "queer/5 and by the fact that the unconscious force attained
control of the executive apparatus but did not gain access to conscious-
ness. Thus we see that some of the concepts involved here are close to the
observations, wrhile others are at increasing distance from them.

In clinical inferences, the distance between observations and con-
cepts may seem even greater. The clinician may infer, for instance, that
the slip pertains to the ancient triangle formed by the subject, his older
brother, and father, which was reactivated by the triangular situation
of the subject, treasurer, and chairman. He may even go further and
infer that homoerotic and aggressive drives involved in jealousy are the
unconscious forces which conflict here with the ego's defenses against
them and interfere with ego interests.

No wonder psychologists gained the impression that the relation of
psychoanalytic concepts and theories to observations is distant and
arbitrary. But is this impression accurate? Let us suppose that our sub-
ject volunteers for a free association session, and his associations cluster
around the treasurer and the chairman, rather than around the interests
of the group. Will we then be justified in inferring that the agent behind
the word "queer" is an unconscious force directed toward the treasurer
and the chairman? Let us suppose further that the subject's associations
not only corroborate that this unconscious force is an aggressive drive,
but identify it as being of a jealous-suspicious variety. Let us finally as-
sume that, in the course of these associations, the subject comes to
realize that he actually has had filial feelings toward the chairman and
vague, poorly understood feelings of irritation with the treasurer, akin
to those he used to feel toward his brother, and thereby he specifies that
the unconscious force pertains to the subject-brother-father triangle.

True, in this sequence the concepts (unconscious, unconscious im-
pulse, unconscious hostile impulse, unconscious hostile and libidinal im-
pulse, unconscious hostile and libidinal infantile impulse) are in-
creasingly remote from the slip of the tongue which is the original
observation. But the associations, too, are observations and the in-
creasingly remote concepts are introduced in reference to these additional
observations. Thus, the distance between observations and concepts is
not as great as it seems on first sight. But there still remains a difficulty :
the relationship of each more remote concept to the corresponding ad-
ditional observation presupposes the less remote concepts. For instance,
without assuming that the unawareness of the subject is a referent of the

The Structure of Psychoanalytic Theory 119

descriptive concept unconscious and without assuming the unconsciously
motivated character of the slip, it would make no sense to infer that
the subject's associations specify the pertinence of the aggressive impulse
(indicated by the slip) to the treasurer and the chairman.

This relationship between observations and concepts is common to
all sciences: observations demonstrate theoretical relationships only to
those who already conceive of the observed in terms of the theory's con-
cepts. But the psychologist seems to overlook this truism when it comes
to psychoanalysis. This oversight is so common that the lack of systematic
treatments of the theory alone cannot account for it. There must be
other reasons, and a few of these will be conjectured:

The psychologist is accustomed to explicit — and, indeed, operational
• — definitions of concepts and is wary of psychoanalysis3 definitions of
concepts. He suspects that the mutual implications of its concepts hide
a vicious circle. In the lack of a systematic statement of the theory, we
can sympathize with his wariness, but we must keep in mind that in
physics nobody would think of asking for an explicit definition of energy
that did not involve the concept of work (which in turn involves the
concepts of path and force, which in turn involve mass and acceleration,
which in turn involve time and velocity, which in turn involve space
and time). We shall not dwell here on the demand for operational
definitions: Feigl [71] and Frenkel-Brunswik [92] have demonstrated
that in this matter psychologists have tried to be more Catholic than the
Pope, and that operational definitions of all its concepts have never been
demanded of any science.

The clinical psychoanalyst is deft and nonchalant in using concepts
at a great distance from the observations. For instance, he may con-
jecture from the word "queer" what might be involved in this slip, by-
passing the intervening observations (e.g., associations) and concepts. It
may be a well-supported conjecture, if the patient's previous productions
converge on it; or it may be a poorly supported one, if the analyst is
more imaginative than careful. It may even help the patient to insight
if it is conveyed to him. But a conjecture it remains until the patient's as-
sociations or other productions confirm it. Some such conjectures are
supported by so much experience, and pertain to relationships so
common, that they are almost certain. These are particularly prone to
turn into cliches, to give the outsider the impression of arbitrariness or
of an uncanny "second sight," and to oversimplify the complexity of
the theoretical relationships even in the psychoanalyst's mind. Actually
the psychoanalyst's use of these may not differ from an electrician's use
of technical terms and repair- or construction-procedures without his
referring to or even being aware of their theoretical implications. When
the rules of thumb of clinical psychoanalysis are equated with the theory

120 DAVID RAPAPORT

of psychoanalysis, the observations and concepts which bridge the gap
between the basic concepts and the initial observations are inevitably
overlooked.

The psychoanalytic writer and practitioner is inclined to speak of
psychoanalytic concepts and theories in terms of "content." The con-
tent of the word "queer" may serve as an example. The subject's use
of this word is conducive (or seductive) to the conclusion that a homo-
sexual impulse may be involved in the production of this slip. The
content of any slip may suggest the nature of the unconscious motivation
involved in it. Content is an important guide to the practitioner. Many
sensitive and experienced psychoanalysts are to a great extent guided by
the content of communications. Others are guided by the tone of voice
or other expressions of emotion. The majority of the contributions to
the literature tend to dwell on content to the neglect of other guides.
What is lost sight of — and the practitioner need not necessarily keep this
in focus or even in sight, but those interested in the theory must — is the
functional (and thus also conceptual) relationship to which the content
is a guide. The word "queer" is — by the circumstances of its utterance —
a compromise formation between id motivations and ego controls. This
is one of the functional relations involved in this slip. This slip's content
suggests some of the (aggressive and homoerotic) impulses involved in
this functional relation. But these suggestions make theoretical sense only
if the word "queer" and the circumstances of its utterance are assumed
to be referents of the descriptive concept of the unconscious, of the con-
cept of unconscious motivation, of unconscious drive motivation, of un-
conscious conflict, and of resolution by compromise. No content yields its
full meaning unless its formal characteristics, and those of the time, locus,
and context of its appearance, are taken into consideration, that is to say,
abstracted. The content of the word "queer," and of any communication,
is predictive only in so far as it belongs to the network of the ideas which
represent the drives involved. In other words, the content is a guide be-
cause it belongs to a network of formal relationships. Whether it is the
Oedipus complex, or the castration complex, or an anal fixation, or a
homosexual impulse to which the content refers, it does so always by
virtue of a formal, conceptual relationship. The stress on content seems
to be one of the main causes for overlooking the relationship between
concepts and observables. Psychoanalysts are not the only ones who
make a direct jump from content to unconscious motivation : Rorschach
testers and other projective test "experts" do it too, often with less ex-
perience and always with less collateral information to go by.76 Recently
McClelland [226] called on psychologists to revive their interest in

76 See, however, Schafer's [298] treatment of content as a guide to formal
relationships.

The Structure of Psychoanalytic Theory 121

content. McClelland is right, it is high time to begin the serious experi-
mental study of contents. But a warning of the pitfalls is in place. The
more familiar the contents dealt with by psychoanalytic theory become,
the greater the temptation to bypass and to becloud the conceptual
relations.77

IV. CONSTRUCTION OF FUNCTION FORMS

Psychoanalysis, as a theory, did not make a formal study of the
construction of its functional relationships. Thus whatever can be said
about these must be inferred. The preceding sections covered this ground
as much as seemed feasible. Here we can add only a discussion of the
theory's "function form" in relation to the Lewinian and S-R function
forms.

Lewin's [219] basic function form is B = f(P,E): behavior is a
function both of the environment and the person. Here B represents be-
havior at large and not any specified aspect of it; E represents the en-
vironment as the person's life space at large and not any specified part
of it; and P represents the structural and tensional characteristics of the
intrapersonal regions at the time of the behavior, rather than the person
as a changing historical entity.

B = f(P,E] can be made to take on extreme values which transform
it into the basic S-R function form, if we make three assumptions : First,
there are environmental situations in which behavior (response) is in-
variant in respect to individual differences and intrapersonal changes;
then for all the E values for which this assumption holds, the function
changes into 5 = /(£). This situation is one in which the causal texture
of the environment has become compelling. In terms of psychoanalytic
ego psychology, this is a situation in which no autonomy from external
reality obtains. Second, the E (life space) does not vary from subject to
subject (such variations are not accounted for directly by this equation).
Third, certain behaviors are determined not by the E in general but by

7T Freud seems to have expressed this as follows, responding in a letter to
Abraham on the latter3 s comments on "Mourning and Melancholia'9 [120]: "...
you do not emphasize enough the essential part of my hypothesis, i.e., the topo-
graphical consideration in it, the regression of the libido and the abandoning of
the unconscious cathexis, and that instead you put sadism and anal-erotism in the
foreground as the final explanation. Although you are correct in that, you pass by
the real explanation. Anal-erotism, castration complexes, etc. are ubiquitous sources
of excitation which must have their share in every clinical picture. One time this
is made from them, another time that. Naturally we have the task of ascertaining
what is made from them, but the explanation of the disorder can only be found
in the mechanism — considered dynamically, topographically and economically"
[193, vol. 2, p. 329].

122 DAVID RAPAPORT

a specific element of it; that is to say, B is invariant in respect to E ex-
cept for its single, specified element S. If we then use the term re-
sponse— R — for the so specified behavior, and the term stimulus — S — for
the so specified elements of E, we arrive at the equation R = f(S).

Similarly, B = f(PyE) can be transformed into the function form
of that phase of psychoanalysis in which the role of reality was negligible,
and the role of the person's drives and defenses paramount, in determin-
ing behavior. Embarking on such a transformation we must note that
Lewin, in failing to establish in principle the relations between "geo-
graphic" and "hodological" (life) space, left the door wide open for all
those who wish to transform E into a function of P and thus to turn the
B = i(P,E] equation into B — f(P)Js Werner and Wapner's "sensory-
tonic" theory [323] — by postulating a tonic factor in perception — intro-
duces P into the perceived E, and thus also leaves an opening (however
narrow) for such a transformation. Others, particularly Brunswik,
Heider, Gibson, and recently Klein [197], explicity refused to do so.
x\ccording to Piaget's studies, the "construction of reality" is an onto-
genetic achievement and not a process of "imitative learning" or "con-
ditioning." Thus in his theory, too, genetic considerations can always
resolve E into E ~g(P}) though here all such g functions are actually
of the form En = gn(En-i,P}, and P itself is subject to historical change
[P» = An(P«-i3£n-i)]. Nevertheless, every genetic theory tends to trans-
form B — f(P,E) into B = /(P) : in genetic theories, genetic reduction-
ism is always a temptation.79

Can any dynamic psychology escape such a reduction? Allport's
personalistic psychology has perhaps the most explicit safeguards against
such a reduction, which turns man either into a mechanism ultimately at
the mercy of its environment, or into a solipsistic creature ultimately at
the mercy of his drives. Allport's [6, 8] safeguard against both these
alternatives is expressed in his concept of functional autonomy, which
implies that whatever the genetic (maturational or learning) history of
a function, it may attain autonomy so that it can serve as an irreducible
basis of behavior. In psychoanalytic theory, a similar solution was in-
dependently reached by Hartmann and expressed in his concepts of the
conflict-free sphere and autonomous ego functions. However, it should
be noted that his is a concept of relative autonomy : functions and struc-
tures have only limited autonomy from the drive or learning process
from which they arose; for instance, behaviors determined by such struc-
tures may be over determined by drives; they may be used by drives as

78 Brunswik and Heider point out that Lewin's environment is "encapsulated.55
G. S. Klein points out that Lewin disregarded the "inherent structure33 of the en-
vironment and centered exclusively on its perceived structure.

79 But see Erikson [61, 62, 66].

The Structure of Psychoanalytic Theory 123

means; and under stress their autonomy does not necessarily hold and
they may yield to ontogenetically earlier forms.80 The implications of
Piaget's [254] genetic theory of intelligence are similar: the new and
higher-level "circular reactions15 — and the "schemata" corresponding to
them — attain independence from the schemata from which they arose.
But the lower-order schemata and circular reactions are not replaced
by the higher ones and may always be reactivated when the latter do not
provide the means of coping with the situation encountered.

B = i(PJ£] implies that no broadly valid relationships of the
B = f(P) or the B = f(E] type are possible. If the B = f(P] function
form is to attain validity for more than a narrowly specified range, E
must be introduced into it as an intervening variable. The same consider-
ation holds for the validity of B = f(E), and demands the introduction
of P as an intervening variable. But neither E nor P is a simple variable :
both are functions of other variables, which can be held constant only
for a narrow range of conditions. If we take it for granted that the
person's experience of his life space is our basic information about his
geographic environment, then £ is a function of the geographic environ-
ment and P; and if P changes — as it does — with experience, then it is
a function of preceding P's and E's.

Let us approach the problem from another angle. The extreme values
which make B = f(P,E) go into either B = f(E] or B = /(P) imply
that certain one-to-one relations between stimulus and response, drive
and behavior are possible. But we know empirically that by and large
this is an untenable assumption, since the single S (stimulus) is hardly
ever the only effective factor in E and the single D (drive) is hardly
ever the only effective factor in P. So when S or D is chosen as the in-
dependent variable, the other factors come into play as intervening
variables. In other words, the remarkable thing about human behavior
is that man often meets diverse stimuli by the same behavior, and
identical stimuli often elicit diverse responses. Likewise with motivations :
the same motive may be expressed by a wide variety of behaviors or
satisfied by a variety of objects, and a great variety of motives may be ex-
pressed by the same behavior or satisfied by the same object [see Frenkel-
Brunswik, 91, and Gill, 146]. Therefore, if stimuli or motives are used
as independent variables, it becomes necessary to introduce intervening
variables to account for the flux of the dependent variable. Thus, learn-
ing theory introduced sets, attitudes, etc., as intervening variables, to save
the R = j(S] function form. Where P at large is the independent
variable, E will serve as the intervening one, and vice versa. In psy-

80 This is only a possible consequence of stress, not a necessary one. Among
others, Jacobson [189] and Bond [30] report observations of increased autonomy
and efficiency under stress. See also [280].

124 DAVID RAPAPORT

chologies where the P is differentiated, and the motive (e.g., the drive)
is the independent variable, structures (defenses, controls, etc.) will
appear as intervening variables. In those psychologies wrhere the E is
differentiated, context and "setting" will appear as intervening variables.
In conclusion: by the very nature of psychoanalytic theory, inter-
vening variables are indispensable in its function forms. It would seem
that this holds true for all dynamic psychologies, and the range of inter-
vening variables they will use will depend upon the degree of autonomy
they assign to the structures and functions involved in the phenomena
studied.81

V. THE PROBLEM OF QUANTIFICATION

It would be simplest to restate at this point that psychoanalysis as a
discipline has not attempted quantification, and avoid the whole issue.
One could express justified impatience with the "furor of measuring"
which has gripped psychology, partly by referring to the history of
evolution theory, in which precious little was measured for a long while,
and partly by referring to how much is being feverishly measured in
psychological laboratories without good reason. Sometimes one has the
impression that the hope in such measuring is well expressed in the
Hungarian proverb, "Even the blind hen does at times peck a grain."

But the issue of quantification cannot be dismissed lightly. Psycho-
analysis— like all other sciences — orders, equates, compares, and dis-
tinguishes observables, and these procedures, once made precise, reveal
themselves as mathematical operations [cf. Piaget, 252, vol. 1]. Thus all
sciences, in striving to make their assertions precise, move toward a
mathematization of the relationships they establish by their procedures.
Since mathematization may be either metric or nonmetric, quantifica-
tion is only one form of it. In contrast to the customary quantification,

81 It can be argued that the use of intervening variables does not depend on the
degree of autonomy the system assigns to structures and functions, but rather on
the observational method by which the data the theory accounts for are obtained.
For example, R. R. Holt (personal communication) comments: "Skinner does not
need intervening variables because he forces behavior into a narrow compass
where S and R can be directly related mathematically." It is probable, however,
that Skinner's procedure, too, is rooted in a choice of autonomous (automatized)
relationships. It would seem that the degree of autonomy, as well as the role
assigned to intervening variables, depends on the observational method used. The
psychoanalytic method alone scarcely allowed, and certainly did not require, the
theory to introduce the concept of autonomy. It was introduced when data ob-
tained by other observational methods were also considered by the theory. Con-
versely, Hebb [169] seems to have realized, when he considered methods of ob-
servation (e.g., Senden's) other than conditioning, that the S-R relation is not
free (autonomous) from what passes between the S and the R.

The Structure of Psychoanalytic Theory 125

Lewin [217] attempted to introduce topology and Piaget [253] to intro-
duce group theory into psychology as nonmetric mathematizations. Non-
metric mathematization does not necessarily exclude quantification; in-
deed it is not clear whether Lewin's quantifications are the result of his
success or failure in nonmetric mathematization. The problem of metric
mathematization is what kind of quantification, if any, is appropriate to
what kind of psychological relationship.

It is probable that psychoanalysis has not developed a mathematiza-
tion, and that academic psychology has not gotten far with its strenuous
efforts at quantification, because they are both sciences in an early phase
of development. Whether it is metric or nonmetric mathematization that
psychoanalysis is headed for is hard even to discuss at this point. It would
certainly be premature to judge that quantification is the kind of mathe-
matization which is appropriate to psychoanalytic theory. The following
discussion of quantification implies no such judgment, but merely this
attitude: since the question of quantification has been raised, and since
quantification may prove to be the mathematization appropriate to psy-
choanalysis, some of the problems it involves should be explored. This
section will dwell on two topics: on the quasi-quantitative concept of
cathexis, which of all the concepts of the theory seems to call most
urgently for quantification, and on the kind of quantification — if any —
required by the theory.

A. Cathexes82

The psychoanalytic theory contains quasi-quantitative concepts. The
most conspicuous of these are the drives, which are conceived of as
forces, and the cathexes they expend, which are conceived of as quantities
of energy. Why then have these not been measured? To answer this
question it is necessary to discuss the distinctions psychoanalytic theory
makes between various forms of energy.

1. The muscular energy of behavior is not the psychological energy
that psychoanalytic theory speaks of: the psychological forces which in
their work expend psychological energy only release the forces that ex-
pend the biochemical energy of muscles.

2. Psychological energy (in the main) is considered as of drive origin,
and to account for its major forms of manifestation, two transformation
processes are postulated : binding and neutralization. Both of these result
in forms of energy (bound, neutralized] which differ from the original
(mobile] form of drive energy.

3. These three forms of energy, and the two major processes of trans-
formation, may be characterized as follows :

82 For references, see pp. 91-93, 113-114, above.

126 DAVID RAPAPORT

a. Mobile energy abides by the pleasure principle: It tends toward
Immediate discharge by the shortest route, and brooks neither delay nor
detour. These direct discharges may take the form of action, idea, or
affect (e.g. impulse action, compulsive ritual, random action; delusion,
obsessional idea; and uncontrolled affect storm, such as a tantrum or
panic, etc.). The processes which expend mobile energy are con-
ceptualized as primary processes. They make use of several mechanisms
(i.e., specific transformations of energy) as, for instance, condensation,
displacement, substitution, symbolization, etc. These mechanisms come
into play in all facets of behavior, though they are commonly illustrated
by examples taken from ideation (e.g., dreams). Mobile energy, though
its function is particularly well illustrated by the behaviors mentioned
above, does not refer to a class of behaviors but to a component of all
behavior.

b. Bound energy is defined as energy tied up in structures. Breuer
[35, pp. 140-141] compared it with the tonic innervation of muscles.
The structures, the building of which amounts to a binding of energy,
are conceptualized on the one hand as those controlling and defensive
structures of the ego which make ordered thought as well as controlled
affect and goal-directed behavior possible, and on the other hand as those
which are the means (information; habits; concepts; anticipatory, gram-
matical, syntactic, and logical patterns, etc.) used by ordered thought,
controlled affect, and goal-directed behavior. The processes made pos-
sible by these defensive and controlling structures, and by these structures
of means character, are conceptualized as secondary processes. The trans-
formation of binding changes mobile energies into bound energies. The
structures thus created counteract the mobility of unbound energies, and
also serve as the means (apparatuses) by which the latter are expended
and controlled. Compared with the great energy expenditure in primary
processes, the structures formed by binding can function (autonomously)
with a minimal expenditure of psychological energy, and by controlling
the discharge of mobile (great intensity) energies they create high
potentials for action. Like physical mechanisms, they transform, save,
and expend energy. The concepts of the binding process and of the
structures which it creates account for that aspect of the psychological
organization which does not reduce, but maintains or even increases,
tension [see Freud, 98, pp. 533-534; and Allport, 8].

£. Neutralized energy is defined as energy whose tendency to follow
the pleasure principle (direct immediate discharge) is decreased. This
definition implies a spectrum of energy forms, ranging from barely
neutralized to highly neutralized energies. The process of neutralization
is defined as the transformation by which drive energies, the ideal type
of which is considered nonneutralized, are transformed into energies of

The Structure of Psychoanalytic Theory 127

various degrees of neutralization. The discharge (entropic) tendency is
common to all energy (physical as well as psychological) : how can we
conceive of energies which abide by it only more or less? The limitations
of our systematic knowledge permit only an answer by analogy:

The entropy principle certainly obtains for closed systems of physical
energy, but organisms, like other open systems and like man-made
mechanical structures, postpone and obstruct the operation of the
entropic tendency. Organisms do this by being structured and by build-
ing further structures [cf. Schroedinger, 305]. Man-made structures do it
by preventing expenditures of kinetic energy (as in dams) and thus
transforming it into potential energy, and by controlling the expenditure
of kinetic energy with small variations of potential energy (as on the grid
of the electronic tube) .

Neutralization is considered to be the result of structure building
by the process of binding [see 274, and 268, particularly part 7]. These
structures, by raising the discharge thresholds of drive energies and by
building new controlling "dams," obstruct the tendency toward direct
discharge, enforce delay and detour, and thus give rise to derivative moti-
vations whose tendency toward immediate and direct discharge is de-
creased : thus, a step toward the neutralization of cathexes is made. With
further structure building, further derivative motives arise, which expend
cathexes of an even higher degree of neutralization.

Observations also necessitate the assumption of transformations which
reverse the effects of binding and neutralization. These observations per-
tain on the one hand to the weakening of controls and defenses, and on
the other to the so-called libidinization or aggressivization of functions
and structures. These transformations may be termed "mobilization35 or
"deneutralization."83 Referents of these transformations may be observed
in special normal states (e.g., dreams), in pathological conditions (e.g.,
compulsions and delusions), etc.

The complexity of this theory of psychological energies and of their
relationship to the motor energy of behavior has far-reaching conse-
quences for quantification. The motor energy of behavior is "controlled"
and "released" by the economics of psychological energies and by the
corresponding dynamics of the psychological forces which operate through
psychological structures. It might be suggested that this relationship is
akin to the control of large amounts of energy (muscular) by an in-
formation network operating with smaller amounts of energy (psy-

83 They occur in the process of regression. Freud discusses them as the dissolu-
tion by regression of the fusion of libidinal and aggressive drives (defusion) [see
131, pp. 46-48; see also 148, chapter on "The Metapsychology of Hypnosis and
Regression"].

128 DAVID RAPAPORT

chological). In turn, within the range of psychological energies we find a
similar relationship : the expenditure of large amounts of energy ( mobile
drive energy) is controlled by a network operating with smaller amounts
of energy (bound and neutralized energy) . Moreover, since the processes
of binding (structure building) and neutralization recur, creating ever
new layers of the ascending hierarchy of psychological organization, we
are faced with a whole array of controlling networks arranged in depth.

One of the obstacles to quantification now becomes obvious. Overt
behavior is as a rule a remote representation of the psychological
processes which give rise to it. Thus, even though observations strongly
suggest the need for concepts like cathexis and those referring to cathectic
transformations, the measurement of their referents is made difficult by
the circumstance that the same set of observations necessitates the as-
sumption of a complex hierarchical control organization regulating
cathectic expenditure and transformation by means of structures operat-
ing with small amounts of cathexis. If the theory were systematically
tight, its definitions explicit, and its implicative rules specific, the dis-
continuities, resulting from the multiple controls which cathectic ex-
penditures are subject to, would not obstruct quantification. But the
theory is far from being that tightly knit. The best index of the theory's
looseness is that the volume of its experimentally verified propositions
would be ample to confirm a tighter theory.

Yet the situation is not as hopeless as the complexities described
suggest. The theory of cathexes does include quasi-quantitative proposi-
tions in the form of inequalities. For instance, the following inequality
holds for mobile cathexes: in drive action, the quantity of cathexis is
greater than that in affect charge, which in turn is greater than that in
an idea. Such a series of inequalities is per se a primitive (intensive)
quantification84 and this kind of quantification is inherent in the theory7.
For instance, there is no reason why the degrees of mobility of cathexis
(or conversely, the degrees of neutralization) should not be expressed in
terms of such inequalities. Indeed, R. R. Holt's [178] study of primary
process manifestations in the Rorschach test did just that. Such ordinal
scaling, using the psychologist's rating procedures, seems for the present
the quantification method of choice for the primitive quantitative rela-
tionships of the theory. Some of its difficulties, however, should be men-
tioned here :

Ordinal scaling of primary process phenomena may distract attention
from the fact that the theory does not posit a simple continuum of
neutralization. The decrease of mobility goes along with binding (struc-

84 Cf. Piaget's [252] discussion of intensive quantification. Altogether, Piaget's
discussions of the development of quantity concepts and its relation to logic and
mathematics are relevant to the quantification problems of psychology.

The Structure of Psychoanalytic Theory 129

ture-building), that is, with the establishment of new hierarchic levels
which differ from each other not only in the degree of mobility of
cathexis, but also in their structures and in their kinds of motivations.
While the degree of mobility remains a common parameter through-
out the various hierarchic levels, the qualitative differences in structures
and motivations from level to level make it difficult to find that feature
of behavior which, when rated and scaled, will quantify that common
parameter. Whenever the behavior feature chosen for rating is not
appropriate, apples and pears will be compared. R. R. Holt seems to
have avoided this pitfall, but it required a considerable mastery of the
theory to do so, and the measures used remained gross.

Thus, individual instances of primary process phenomena do not
offer an insurmountable obstacle to quantification. But how about a be-
havior segment (a Rorschach or a TAT record, or a clinical interview)
which contains several such instances? Each of these can be rated. But
may we count them? If we do, what is our justification for doing so?
If we decide to weight them, are the weights additive? These questions are
not yet answered. We do not even know where the answers will come
from. We may have to accept purely empirical, theoretically unsupported
answers for a long while, in the hope that these will show us that the
theory7 has (or can be expanded to have) the answers. It is also possible
that the empirical answers will radically change the theory. This problem
is not specific to Holt's study. In food deprivation studies using TAT
stories, we find individual differences in the stories of a group of equally
deprived subjects: some stories contain much material distantly related
to food, others contain little food-related material, but what there is, is
closely related. Can the ratings of these individually differing products
be added up? The relationship between drive intensity (amount of
cathexis) on the one hand and the frequency and intensity of its in-
dicators on the other is a significant unsolved problem of quantification.

B. Dimensional Quantification85

What are the general prospects for the quantification of the variables
of this theory?

Before attempting an answer to this question, let us state that the
urgent tasks of this theory are in the relationships it posits which re-
quire systematization, and in the areas which require new observations.
Without stressing that, among other things, much of ego psychology is
still uncharted territory, and that our knowledge of affects is in urgent
need of systematization, etc., the very discussion of quantification may

85 Since the completion of this manuscript, A. Menkes and J. Menkes have
published a paper [231] which contains an example of this kind of quantification
and goes a considerable way in demonstrating the necessity for such.

130 DAVID RAPAPORT

misrepresent the actual situation: mathematization in general and
quantification in particular require a systematized and tightly knit theory.

When the physicist measures, he knows the dimensions of his ob-
servables as expressed in terms of the CGS (centimeter, gram, second)
system, and when he establishes a constant he knows that its dimension
is such as to make his equation not only quantitatively but also di-
mensionally true. In s = (g/2)t2 the dimension of s is C, of t is S, and
of g is C/S2; thus, substituting these dimensions, we get C = (C/S2}S2,
indicating that the equation is dirnensionally true. The classic scale of
hardness is a means of quantification too. But instead of a dimensional
measure, it provides only an ad hoc quantification. Most — if not all —
measurements (e.g., IQ's) of present-day psychology are ad hoc quanti-
fications. Without a systematized theory, no dimensional quantification is
possible. In physics, nobody would try to test a theory by a measurement
without first ascertaining the relevance of what he measures and how he
measures it. The dimensions are the criteria of relevance. Psychologists,
however, "test" psychoanalytic propositions without studying and system-
atizing the theory which gives meaning to these propositions. Theories
can be tested only when they are taken seriously. To test is to mathema-
tize and to mathematize is to discover, in the relationships posited by the
theory, relationships of a higher order of abstraction. Such abstractions
cannot be derived from isolated propositions, but only from the system
of relationships which link these to each other.

So far we do not know how to achieve a dimensional quantification
of psychoanalytic variables; and yet we cannot sit with folded hands,
since additional observations are needed for the systematization of the
theory and for dimensional quantification. Thus in gathering new ob-
servations we must be satisfied with ad hoc quantifications, but we must
not lose sight of the goal of dimensional quantification. To achieve that,
we will have to learn to consider the locus of our variables in the motiva-
tional and structural hierarchy and to play variables against each other
so as to arrive at equations which represent actual balances of forces, or
balances between structures and forces, etc. Progress toward dimensional
quantification will at every step require long series of experiments which
vary the experimental conditions systematically. The currently fashion-
able one-shot experiments (probably fostered both by the premium put
on publication and by the publication policy of psychological journals)
militate against progress toward dimensional quantification. One-shot
experiments, naturally enough, use ad hoc quantifications, and only
rarely cogwheel into the ad hoc quantifications of other experiments.
Lewinian experiments in affect and action psychology avoided this pitfall
to some extent and showed how ad hoc relationships can be avoided by
systematic variation of experimental conditions directed by a cohesive

The Structure of Psychoanalytic Theory 131

theory. But the reports of these experiments are In German and thus
have been little read, except in Ellis's "52] excerpts or Lewin's [216]
summary, neither of which conveys the method.

One of the banes of ad hoc quantification is that even when it yields
statistically reliable results, these may be due to sheer luck in choosing
the experimental tasks and subjects. Even the apparently precise replica-
tion of an experiment may bring different results. The crucial dimensions
not being known, unnoticed "minor33 variations of the setup affect the
results. In other words, without knowing the dimensions involved it is
impossible to predict what changes will make for hierarchic differences
and for what types of subjects will the objectively "precise" replication
amount to a radically different setup. George Klein [197, 200] has shown
that something of this sort was involved in the "now you see it, now you
don't53 character of the Bruner-Goodman [37] effect.

Now, as to the possibility and prerequisites of dimensional quanti-
fication: First, dimensional quantification in psychology may not be
feasible. We would be reluctant to entertain this possibility, partly be-
cause it would require negative proof, which is difficult if not impossible
to obtain, and partly because it would discourage further research.
Second, the quest for dimensional quantification may lead to a nonrnetric
mathematization. Third, a dimensional quantification may develop.

The avenue by which we may be able to arrive at a choice among
these three possibilities will — to my mind — be paved by a new start on
the problem of learning:

The physical dimensions are mass, space, and time. Physics ex-
presses both the movement of mass (i.e., its changes of position in space
and time) and the changes in the structure of mass, as well as the
gradients and causes (force, energy) of such changes, in terms of these
dimensions. If we were to have psychological dimensions, they too
would have to be able to express psychological processes as well as
psychological structures and their changes. In psychoanalytic theory,
structures play such a crucial role that as long as the propensities
and changes of psychological structure cannot be expressed in the same
dimensions as psychological processes, dimensional quantification is but a
pious hope. In other words, the study of the process of psychological
structure formation seems to be the prime requisite for progress toward
dimensional quantification. We must establish how processes turn into
structures, how a structure, once formed, changes, and how it gives rise
to and influences processes. This could be achieved, for instance, by
studying the processes by which Hebb's hypothetical structures (as-
semblies and phase sequences) are formed and changed, as well as the
processes by which these structures change new ongoing processes. Like-
wise, this could be achieved by the study of those structures whose genesis

132 DAVID RAPAPORT

and function psychoanalysis conceives of as follows: when drives en-
counter an obstacle to the discharge of their cathexes, structures are
formed and these structures thereafter serve both as obstacles to (de-
fenses against) and controls and means of discharge. These examples
refer to changes wrought by experience. Whether or not all structure
formation (in that broad sense which takes account of the epigenetic-
maturational matrix)86 should be considered learning (i.e., abiding
change wrought by experience) is both an empirical and a conceptual
problem. But it seems that all learning may be looked upon as a process
of structure formation. The processes of verbal learning and habit
formation may well be considered subordinate to this broader category,
though their study may or may not be revealing of the relationship be-
tween process and structure.

What study will reveal this relationship? Thirty years ago, Adams [2]
suggested that the main obstacle in the way of the study of the learning
process is its slowness. Hebb [169] pointed to the slow rate of early
learning processes. The burden of Piaget's [254, 255, 256, etc.] develop-
mental studies in and since The Origins of Intelligence is the same.
It is possible that only longitudinal studies can clarify the relationship
between process and structure. But since the methodology of longitudinal
studies is still obscure, this is a dim hope. Perhaps the answer will come
from a new attack on learning as structure formation, which will take
account of Hebb's assumption that late learning operates by recombining
already established "phase sequences' 3 and will thus center on the changes
in, rather than on the origin of, such phase sequences.

The immediate outlook for an early clarification of the process of
structure formation seems none too rosy. Yet this clarification appears
to be the prerequisite for dimensional quantification in psychoanalysis
in particular, and perhaps even in psychology at large.

But the quest for dimensional quantification must not amount to
a disdain for ad hoc quantification. The latter seems to be a step
toward the former, provided it is clearly understood that ad hoc quanti-
fication itself does not locate hierarchically the structures and functions
which it crudely quantifies. The possibility of arriving at a dimensional
quantification can be kept open by matching the care and ingenuity ex-
pended on ad hoc quantifications with an unremitting alertness for the
hierarchic locus of the relationships so quantified.

All this discussion of quantification is, however, in a sense abstract
and sterile. A proper discussion would have to start out with an analysis
of the experimental literature pertaining to Freudian propositions. We
have several surveys of this literature [e.g., 308, 173], but their concern
is: what psychoanalytic propositions are confirmed by "objective studies?"
86 Cf. pp. 86-88, above.

The Structure of Psychoanalytic Theory 133

A survey which could advance the solution of the mathematization or
quantification problem would have to center not on the results of these
studies but on their method, on the variables which were the targets
of quantification in them, and on the technique of quantification used by
them. Short of a breakthrough by means of experimental ingenuity
coupled with thorough theoretical grounding, such a survey seems to
hold the best promise of progress toward the solution of the quantification
problem.

VI. THE FORMAL ORGANIZATION OF THE SYSTEM

A. The Present Status of the System

The theory of psychoanalysis grew by successive spurts in the fifty
years of Freud's work. Additions and revisions make it appear more
like a patchwork than an architectonic design, since their consequences
for the structure of the system have often remained a matter of a passing
comment by Freud or isolated papers by other psychoanalysts. This in
itself bespeaks a looseness of the theory and its lack of an explicit canon
according to which revisions and additions are to be fitted into its system.
Yet psychoanalytic theory does have an impressive structural unity,
though it is hidden under the layers of progressive additions and modi-
fications, and has not been disentangled and independently stated.

The "revisors" of Freud's theory further obscured its structural unity.
Jung and Adler, who created relatively independent theories, failed to
give these a systematic form which could have sharply distinguished them
from Freudian theory. The situation is even worse with the "revisions33 of
Stekel, Rank, Sullivan, Homey, M. Klein, Kardiner, Alexander, French,
Reik, Fromm, Rado [see Munroe, 240] . While each attacked and denied
certain Freudian propositions, and replaced them by others (which
often contained a valid core), none of these authors stated how their re-
visions affect the theory as a whole. Some of them (Stekel, Kardiner,
Alexander, French, Reik) have asserted that their revisions do not affect
the rest of the system, although they made no attempt to demonstrate
this. Others (Rank, Horney, Sullivan, Rado) have implied that Freud's
system has been replaced by their own, although they never presented a
full elaboration of their systems. No Neo-Freudian has taken cognizance
of, and has integrated his own contribution with, the whole of psycho-
analytic theory. Nor is there a single attempt to replace it with a whole
system that demonstrably accounts for all the phenomena psychoanalytic
theory claims to explain. Such an attempt could obviously include a
demonstration that some of the problems psychoanalytic theory dealt
with are pseudoproblems which can be ignored. The lack of an explicit

134 DAVID RAPAPORT

statement of the theory is as much responsible for aH this as are the
"revisors" themselves, who may have felt that they were not obliged to
disentangle the system before they re\ised it. Study of Neo-Freudian
writing often makes one wonder whether the authors were aware of the
existence and nature of the implicit system of psychoanalytic theory.

There are three outstanding rudimentary1 statements of the theory's
system.

First 9 Freud's seventh chapter of The Interpretation of Dreams [98]
and his "Papers on Metapsychology" [108, 110, 114, 115, 116, 117, 119,
120] are attempts to present the system. One of the most puzzling prob-
lems of the history of psychoanalysis is why they were so little noticed.
The fact that the form of these attempts is not systematic does not seem
to explain this fully. The formulations of the present essay derive from
these writings, and so do the other attempts at systematization to be
mentioned here.

Second, FenichePs The Psychoanalytic Theory of Neurosis [73],
while it is focused on the clinical theory of psychoanalysis, does per-
sistently invoke the general theory and thus gives a sense of its system.
Yet the latter remains implicit, and the experimenter who wishes to start
from FenichePs formulations must first disentangle them from their
clinical matrix. With FenichePs death, psychoanalysis lost one of its few
systematizers. His essay on the theory of technique [72] is a beginning of
the systematization of the theory of therapy. His posthumously published
Collected Papers [77] contains systematic discussions of M. Klein, Kaiser,
Fromm, and other "revisionists." These discussions, as well as his paper
on Freud's theory of the death instinct, show that the psychoanalytic
theory is sufficiently cohesive to permit systematic exclusion and in-
clusion of new contributions.

Third, the development of ego psychology is perhaps the clearest
demonstration of the systematic nature of psychoanalytic theory. In
Anna Freud's [93] work the clinical theory of defenses begins to take
a systematic form; in Erikson's [61] work the development of the ego
and the psychosocial theory of psychoanalysis takes shape; and in Hart-
mann's [157] work [complemented by Kris's and Loewenstein's, 167,
168, 206] the theory of the ego develops hand in hand with a progressive
crystallization of the general theory of psychoanalysis. All these con-
tributions show that psychoanalytic theory can grow organically so as to
include the valid observations and formulations of the Neo-Freudians,
without becoming an incoherent patchwork and without the necessity
of discarding any of its major segments. They demonstrate that the
theory has sufficient systematic coherence not only to reject incompatible
solutions, but also to develop compatible theories of the ego, of reality,
of interpersonal relationships, and of social psychology.

The Structure of Psychoanalytic Theory 135

B. The Desirable Level of Formalization

The desirable level of fomialization is, in a sense, an empirical ques-
tion. Since everybody wants to be on the side of the angels, we may as-
sume that reaching maximal explicitness is the ideal and only the limita-
tions of our knowledge stand in the way. Newton's axiomatization was
explicit and its heuristic value shows that it was desirable. But the
systematic and heuristic value of Einstein's last formalizations is much
questioned. Present-day physics has no unified axiomatic system. All in
all, probably only experience can decide when and how far axiomatiza-
tion can be meaningfully pushed in psychology or in any other science.

Yet it may be worth while to raise the question: why are psy-
chologists so concerned with axiomatization? Actually, axiomatization
has always been a late product in every science. Centuries of Egyptian
geometry preceded Euclid. Newton had not only Galileo and Kepler, but
thousands of years of physics behind him. Sciences do not arise from, but
culminate in, axiomatics. Axiomatic systems do not reveal the tracks of a
science's development: they conceal them. They do what so many psy-
chologists do who arrive at their results with great difficulty (like the rest
of us), but from reading their published papers one would never guess
that; they seem to reveal a foresight which puts to shame all others who
deal with human beings or govern human affairs.

Does the yearning for axiomatization mean that psychologists believe
psychology can arrive at its future by lifting itself by its own bootstraps?
Are we really to believe that we can guess our way through to axiomatics
and bypass the long road other sciences have had to travel toward it?
Beat the other sciences at the game? Or just simply profit by their ex-
perience? But what if our guesses lead to a disregard of the empirical
evidence we already have and to a lack of concern for the evidence that
is not yet in? What if the attempts at short-cutting the arduous path of
development lead only to endless detours — much longer than the "long
and hard" empirical route, and futile, to boot? Is it possible that psy-
chologists ignore what the natural scientist [24] and the historian of
science [46] have come to recognize : that scientific discovery starts from
intuition and not from deduction?

This is not to question that psychology can profit by the experience
of other sciences, nor to make light of axiomatics as an ideal, nor to
minimize its importance in the development of sciences, nor to contend
that theory making (including axiomatics) is not as essential to science
as "measurement33: intuition or hunch is theory. The point is that in
present-day psychology the measuring furor seems to have made an
unholy alliance with an axiomatic -furor, and between the two of them
they may well doom psychology to stagnation.

136 DAVID RAPAPORT

Piaget in his Epistemologie Genetique [252] examined both the
history of mathematics and the ontogenetic development of mathematical
reasoning in an attempt to explain how mathematics can be simul-
taneously deductively rigorous and yet fertile. The study of his investiga-
tions is a good antidote to premature axiomatization.

Psychoanalysis is in sore need of systematization, because without
it the experimenter is likely to continue to test isolated and misconstrued
propositions, unaware of their actual theoretical context. But systematiza-
tion is a long way from formalization and axiomatization. Much addi-
tional knowledge will have to accumulate before we can even begin to
work on the latter tasks.

VII. THE RANGE OF THE SYSTEM'S APPLICATIONS

The theory, though it originated in the study of pathology, has
always claimed to explain normal behavior and development also —
Psychopathology of Everyday Life [99], Wit and Its Relation to the Un-
conscious [100], "Humour" [134], Three Essays on the Theory of
Sexuality [101].

Moreover, Freud demonstrated that the theory and its methods
can be fruitfully applied to anthropology and prehistory [Totem and
Taboo, 109]; to the study of literature [Delusion and Dream, 102, "The
Relation of the Poet to Day-dreaming,33 103, "Dostoevsky and Parri-
cide,53 133, "The Theme of the Three Caskets," 1 12] ; to the study of art
[Leonardo da Vinci, 105, "The Moses of Michelangelo," 113]; to the
study of mythology, folklore, and legend [Totem and Taboo, 109, "A
Mythological Parallel to a Visual Obsession," 118, "Medusa's Head,"
125, "The Occurrence in Dreams of Material from Fairy-tales," 111];
to the study of language ["The Antithetical Sense of Primal Words,"
104]; to the study of religion [Totem and Taboo, 109, "A Religious Ex-
perience," 135, The Future of an Illusion, 132, Moses and Monotheism,
141]; to the study of history [Moses and Monotheism, 141]; and to the
study of society [Totem and Taboo, 109, Group Psychology and the
Analysis of the Ego, 124, Civilization and Its Discontents, 136, "Why
War?" 138]. Finally, Freud at various times asserted the applicability
of his method and theory to those phenomena which we subsume under
the term psychosomatics.

Indeed, Freud considered all human behavior and endeavor to be
within the purview of psychoanalysis. Psychoanalysts followed his lead
and their literature abounds in papers and books dealing with the fields
listed. Even though these contributions aroused heated and often
acrimonious debate, and even though their cogency and their scholarship
in the field in question have been criticized sharply and often rightly,

The Structure of Psychoanalytic Theory 137

the present situation in all of these fields seems to bear out Freud's early
claim.

In summary, psychoanalytic theory has asserted an all-inclusive ap-
plicability to the study of man. Psychoanalysts have acted to make this
claim good. Investigators in the various fields approached by psycho-
analysis have adopted some of its methods, concepts, theories, and out-
look. There is sufficient evidence that the claim has a substantially valid
core.

Now we come to the applications of psychoanalysis to psychology
proper. Here we can give only a brief sketch of the complex problems
involved [see Shakow and Rapaport, 309].

Though Freud conceived of psychoanalysis as a general psychology,
little in his theory pertained directly to psychophysics, learning, and per-
ception, the areas central to academic psychology, and he did not at-
tempt to apply his theories or methods to psychology at large.

At first only a few psychoanalysts showed an interest in psychology:
for example, Schilder [299, 300, 301, 302], Bernfeld [13], de Saussure
[297]. But through developmental psychology [Piaget, 247, 248, 249,
250, 251, and Werner, 322], through early clinical-experimental psy-
chology [Murray, 242, 243, and Rosenzweig, 293, 294], through the
influence of projective techniques on clinical psychology [Rorschach, 292,
Morgan and Murray, 238, and others], through learning theory [Dollard
and Miller, 48; HuU, 183; Mowrer, 239], 8T and through psychologists3
growing interest in psychotherapy, psychoanalysis came to exert a power-
ful influence on psychology proper. Most of this influence did not stem
from psychoanalysts' applying their theory and methods to psychology
(Jung, Rorschach, and Murray may be considered exceptions) but
rather from psychologists' attempting to use the conceptions (rather than
the concepts)88 of psychoanalysis.

Only with the development of psychoanalytic ego psychology did
psychoanalysis begin to acquire means for dealing with the usual prob-
lems of psychology, Hartmann [157] then made it explicit that psycho-
analysis is a general psychology, that its interest and application extend
to the field of academic psychology, and proceeded to link psychoanalytic
and psychological propositions to each other. Subsequently, several psy-
choanalysts and psychoanalytically trained psychologists continued to
relate psychological and psychoanalytic conceptions, theories, concepts,

8TFor earlier psychoanalytic influence on learning theory, see E. B. Holt [174,
175], Kempf [196], Humphrey [184, 185], Troland [318], and others.

88 Concepts are terms defined within the framework of a theory, conceptions
are terms and formulations which either precede the definition of the concepts in
the history of a theory or disregard them. Thus statements of conceptions use the
terms of a theory in an imprecise or arbitrary "common-sense" fashion.

138 DAVID RAPAPORT

and methods to each other. It is to be hoped that, as a consequence of
this work, the haphazard "experimental testing" of psychoanalytic
theories and their untested application by clinical psychologists may
eventually give way to their systematic application to psychology7, within
the framework which psychoanalytic ego psychology has begun to build.
To advance this work of mutual application, the theory of psychoanalysis
must face two major tasks besides systematization : coming to terms with
Piagefs theory and developing a learning theory.

If Piagefs [252, 254, 255, 256] findings are confirmed, psycho-
analysis will have to come to terms with his developmental theory as an
indispensable segment of the theory of ego development. The problems
to be solved before this can be accomplished cannot be sketched here.89

Our discussion of learning (see Section V. B.) suggested that
dimensional quantification may not be possible without a prior clarifica-
tion of the process of structure formation and learning. But the solution
of the problem of structure formation may also be one of the pre-
requisites for a unified theory of cognition (including perception), for the
clarification of the methodology of developmental studies, and perhaps
for the solution of still other issues crucial both for the systematic de-
velopment of psychoanalysis and for the mutual and fertile application
of psychological and psychoanalytic methods and theories.

VIII. HISTORY OF THE SYSTEM'S RESEARCH MEDIATION

It would take volumes to sketch and critically appraise all the
research that has been "mediated" by psychoanalytic theory, by hunches
derived from it, questions raised by it, and methods originating in it.
It is not feasible to list even the highlights of such research in the
fields mentioned in the previous section. Therefore, we will restrict our-
selves to a cursory survey of its research mediation in psychodynamics
and psychology.

There are, first of all, the clinical studies which fill the psycho-
analytic as well as the Neo-Freudian literature. Moreover, it has been
a long time since any clinical psychiatric case study could be made with-
out drawing on psychoanalytic theory, which, to a greater or lesser ex-
tent, has mediated much of what passes today for clinical psychiatric re-
search. Finally, the psychosomatic investigations of the last two decades
arose, in the main, from psychoanalytic studies of organ neuroses, were
nursed to a more or less general acceptance by the work of psycho-
analysts like F. Deutsch, Alexander and French, Dunbar, and Binger,
and have been turned into everyday clinical research by the efforts of
Kubie, Kaufman, M. Lewin, Romano, and many others.

89 See P. H. Wolffs [328] study of Piaget's theory and his discussion of its
relation to psychoanalysis.

The Structure of Psychoanalytic Theory 139

A selection from one area of more or less systematic studies which
issued from psychoanalysis proper was collected and reviewed in
Organization and Pathology of Thought [268], but a full survey of all
such studies has not yet been made.

Projective techniques, which have come to play an increasing role
both as subject matter and as tools of psychological research, had their
origins in psychoanalytic theory. Rorschach and Murray were steeped in
psychoanalysis and their tests are informed by psychoanalytic conceptions.
In fact, these tests came into clinical use carried by, and carrying, the im-
pact of psychoanalytic theory; they used segments of that theory for
their rationale and interpretation [176, 281]; and they wrere used to
"test" psychoanalytic propositions. Moreover, both these and play tests
(deriving from the play techniques of therapy) bred a vast array of newT
projective tests, founded on and "testing" further psychoanalytic con-
ceptions. How valid their connection to and their "testing" of psycho-
analytic assumptions were, need not concern us here: "research" was
mediated.

Throughout the last forty years, psychoanalytic theory has led to
an extensive array of experimental studies on the effect of emotions and
motivations on memory [see 258]. Most of these intended to test the psy-
choanalytic theory of repression, but many failed to distinguish this from
hedonistic pleasure-pain theories or from the law of effect, and few if
any were really conversant with it.

A related area of research mediated by psychoanalysis is that of
motivated perception. Murray [242], N. Sanford [295, 296] pioneered,
and Murphy and his pupils [see survey in 241, chap. 15] continued this
line of investigation, which led to the "new look in perception," be-
ginning with Bruner's [37, 38, 39, 40, 41] work and reflected in Blake
and Ramsey's [29] volume. Among these, from the point of view of this
essay, the work of Klein and his associatesS9a stands out. While all these
studies bear the imprint of the interest in motivation aroused in psy-
chology by psychoanalysis [cf. Boring, 31, pp. 693, 713], the Freudian
influence is not always as obvious in them as is the influence of Freud's
motivation theory in the studies of Murray and Sanford, and that of
Freudian motivation and ego theory in the work of Klein and his
associates.

Psychoanalytic theory was also responsible for the reawakening of
interest, during the last twenty years, in the nature of hypnosis and in the
use of hypnosis as an experimental method. M. Erickson's [53, 54, 55],
Farber and Fisher's [68], and Gill and Brenman's [33, 148] hypnotic
work, as well as Fisher's [83] work with waking suggestion, represent
efforts to apply psychoanalytic theory to hypnosis or to use hypnosis as a
means of psychoanalytic exploration.

89a A survey of these will be found in Klein's [198] volume soon to be published

140 DAVID RAPAPORT

Psychoanalysis stimulated and guided more or less directly many
longitudinal and cross-sectional developmental studies (Benjamin, Esca-
lona, Spitz, K. Wolf, and others). This field is so broad that neither a
further listing of investigators nor a bibliography can be given here. But
a reference to Piaget must be made. Piagefs early work (up until The Ori-
gins of Intelligence, 254, in 1935) on autistic thinking and its socializa-
tion in children was strongly influenced by the psychoanalytic theory.90
Piaget's later work is critical of psychoanalysis, but still appears to show
its influence.

Finally, psychoanalysis — for better or for worse — has also mediated
much research along the lines of the learning theories which originated
at Yale. Whatever view one takes of their ultimate pertinence to psycho-
analytic theory, Miller's experimental work [234, 235, 236], Miller and
Bollard's studies [237], and Mowrer's experiments [239], as well as
those of their many students, certainly arose under psychoanalytic in-
fluence.

But this enumeration of major areas of research mediation by psycho-
analysis in psychology disregards, among other things, social psychologi-
cal research (e.g., on authoritarianism) and does not do justice to the
pioneering work of D. Levy [212, 213], Halverson [155, 156], Murray
[243], J. McV. Hunt [186, 187], and many others. Regrettably, the
existing surveys— Sears [307, 308], Rapaport [258], Hilgard [173]—
are either specialized or incomplete. A careful analytic survey of the
pertinent experimental literature would be a formidable undertaking:
the amount of literature on research purporting to be related and on re-
search actually related to psychoanalytic theory is immense. Yet such a
survey is urgently needed. It would be of most use if it were to center
neither on the design of the experiments nor on their results, but rather
on the relation of the methods used to the theory.

IX. THE EVIDENCE FOR THE SYSTEM

A. Current Status of Positive Evidence

The major body of positive evidence for the theory lies in the field
of accumulated clinical observations. The first achievement of the system
was a phenomenological one: it called attention to a vast array of
phenomena and to the relations between them, and for the first time
made these appear meaningful and amenable to rational consideration.
In regard to these phenomena and relationships, the accumulated clinical
evidence is positive and decisive. The situation is different, however, in
regard to the theoretical propositions of the system. While the evidence

90 See the introductions to these volumes and their other references to Freud.

The Structure of Psychoanalytic Theory 141

In respect to these also seems massive and Imposing, the lack of clarifica-
tion as to what constitutes a valid clinical research method leaves unde-
termined the positive evidential weight of the confirming clinical material
In spite of the various discussions [e.g., Brenman and Gill, 34, 147,
Kubie, 209, 210; Benjamin, 11; Escalona, 67, etc.] on the nature oi
clinical research, and in spite of French's [89] extensive attempt tc
exemplify the method, its principles have not yet been expressed in the
form of a canon. Indeed, many psychologists would question whethei
there is or can be any other canon of research than the experimental
Since it is questionable whether there exists such a thing as the experi-
mental canon, these views need not worn- us. Because a canon of clinical
investigation is lacking, much of the evidence for the theory remains
phenornenological and anecdotal, even if its obviousness and bulk tend tc
lend it a semblance of objective validity. This makes it urgent to rein-
vestigate Freud's case studies with the aim of clarifying whether or not
they can yield a canon of clinical research at the present stage of oui
knowledge.

In the lack of a canon for clinical research, it is difficult to accept as
positive evidence observations which must first be interpreted before it
becomes clear whether or not they confirm the predictions of the theory :
we must be wary lest we smuggle in the confirmation through the inter-
pretation. Axiomatization and/or a canon of investigation protect other
sciences from such circularity. The lack of such safeguards is a real
handicap for this theory, since by the very nature of the relation between
observations and theory, only observations pertaining to basic concepts
and theorems can be free of interpretation (cf. pp. 116-121). For in-
stance, one of the major propositions of the psychoanalytic theory, con-
firmed by observations, is that there are two kinds of mental processes:
primary and secondary. Little or no interpretation of the observations is
needed to demonstrate that pathological, dream, or drug states bring tc
the fore mental processes which do not abide by the laws of orderec
logical thought. But only on this low level of abstraction is the evidence
conclusive without interpretation. As soon as the evidence for the
mechanisms of the primary process is tackled, observation and interpreta-
tion begin to shade into each other. Per se, that should not invalidate th<
evidence, since no science can get along without interpreting its findings
Yet in psychoanalysis the difficulty is that the canon of interpretatioi
itself is in question — or at least not beyond question — and it is likely t<
remain so until the nature of the clinical method has been clarified, o
until experimental methods have been found which provide an inde
pendent base for the theory. As things stand, there is no canon whereb;
valid interpretation can be distinguished from speculation, though 0,
post facto the experienced clinician can distinguish them rather well.

142 DAVID RAPAPORT

We have a few experiments which are free of this difficulty. The ex-
periments on dream symbolism [306, 288] and the related observations
of Silberer [310; see also Rapaport, 276] are the outstanding ones. But
these experiments remain phenomenological in that they demonstrate the
existence of symbolization, rather than the specific conditions of its
occurrence. The Poetzl [257] experiment and Fisher's [84] replication of
it, as impressive as they are, involve interpretation.91

Most of the experimental evidence for the theory is questionable,
even if Sears5 survey [308], which was loaded with negative bias, was in-
clined to accept some of it as positive evidence. The overwhelming ma-
jority of experiments designed to test psychoanalytic propositions display
a blatant lack of interest in the meaning, within the theory of psycho-
analysis, of the propositions tested. Thus most of them certainly did not
measure what they purported to; as for the rest, it is unclear whether or
not they did. Even where the findings appear to confirm a relationship
posited by psychoanalysis, the experiments usually tested only an anal-
ogous relationship on a high level of the hierarchy of psychological or-
ganization. It is not that all these experiments are useless as confirming
evidence, but rather that at this stage of our knowledge it is not clear
what — if anything — they confirm. It is hard to share Hilgard's [173]
enthusiasm for most of the experiments he considers relevant and con-
firming. It is likely that some of the experimental findings will fall into
place when ego psychology has clarified the hierarchic relationships
which obtain in psychological organization. Command of the theory
should help toward making the results of future experiments unequivocal,
but it is not as much of a guarantee of success as ignorance of the theory
is of failure. The experimental psychologist who enters the precincts of
psychodynamics meets the same complexities which the clinical ob-
server has been struggling with for over six decades. There are no "easy
pickings" and the "experimental method" has no magic here.

In conclusion: the extensive experimental evidence for the system,
which would seem to confirm it in terms of the usual criteria of psycho-
logical experiments, cannot be considered conclusive in terms of the
psychoanalytic theory, since most of the experiments disregard the the-
ory's definitions. The extensive clinical evidence, which would seem con-
clusive in terms of the system's internal consistency, fails to be con-
clusive in terms of the usual criteria of science, because there is no estab-

91 1 am not listing here D. Levy's and J. McV. Hunt's experiments, nor others
akin to them, because they are animal experiments, and represent conditions of a
simplicity which does not obtain in man. — By the time of proofreading I had an
opportunity to read the manuscript of the joint study by C. Fisher and I. Paul
presented at the 1958 meetings of the American Psychological Association. It goes
a long way toward meeting the difficulties discussed here.

The Structure of Psychoanalytic Theory 143

lished canon for the interpretation^ of clinical observations. Thus, only
a few observations and experiments (themselves in need of replication)
offer evidence acceptable both in terms of the theory and in terms of
psychology at large.

B. Major Sources of Incompatible Data

It is often assumed that the data and theories of the "dissident"
schools of psychoanalysis [cf . Munroe, 240] are incompatible with and an
embarrassment to psychoanalytic theory. This does not seem to be the
case. Recent developments [for instance, Zetzel, 330, Erikson, 66] in
psychoanalytic theory in general, and in ego psychology in particular,
seem to show that this theory has the foundations for concepts and propo-
sitions which can account for the observations made and the valid rela-
tionships posited by the dissident schools. Thus we find no source of
embarrassment here, but rather a task to be accomplished.

It is at times assumed — particularly by psychologists — that the find-
ings and therapeutic results of Rogers' client-centered therapy are a
source of embarrassment for the psychoanalytic theory. But this is hardly
the case. Rogers3 counseling procedure, at least to begin with, had no
general psychological theory, nor even a theory of personality. The vague
outlines of a theory of personality, which it has developed since then,
seem to form a segment of an ego psychology. Thus the possibility of
contradiction and embarrassment is limited to begin with, and is further
minimized by two other aspects of nondirective counseling. First, the very
idea of nondirectiveness is one of the implications of psychoanalytic
therapy. The method of free association and the analyst's "evenly hover-
ing" [106] attention imply it. They both demand that the patient's prob-
lems not be prejudged and that reliance be placed on his ability to meet
his problems spontaneously. Rogers' criticism of psychoanalysis is well
founded in so far as it implies that with the accumulation of psycho-
analytic knowledge, and with experience in wielding the tools of inter-
pretation, often little room was left for the patient's spontaneity, and too
often the therapist came to be always right and the patient always wrong.
Indeed, psychoanalytic ego psychology, too, may be considered to be a
reaction to this danger, and the emphasis on the activity of the ego as a
crux of therapy seems to have a central place in its therapeutic and gen-
eral theory.93 But the roots of this danger are in the practice rather than
in the theory even of "classical" psychoanalysis. Second, nondirectiveness
is but one aspect of the technique of psychoanalysis, and can be no more

82 The nonexistent scientific canon of interpretation is not to be mistaken for
the well-established clinical techniques of interpretation.

S3Cf. Sullivan, Homey; also P. Bergman [12], Gill [145], and Rapaport [266,
275, 280].

144 DAVID RAPAPORT

than one aspect of any other therapy. Experience has confronted non-
directive counselors with the problems of transference and resistance
familiar to open-eyed therapists of any persuasion. When the "nondirec-
tive approach" faces these problems, it will meet the eternal struggle of
man's spontaneity, goodness, readiness and ability to help himself, against
man's inertia, fear of his own spontaneity, need for help, etc. While it
is true that treating man as a helpless, inert, and needful creature is prone
to demobilize his spontaneity and ability to help himself, it is also true
that man's helplessness, inertia, and need for support will not be elim-
inated by denying that they exist. Therapies or therapists who practice
either sort of denial end up by establishing their own McCarran Act:
sooner or later they announce that this or that kind of patient is not
the right kind for their kind of therapy. Not rarely they go further
and announce that this or that kind of patient is "not treatable." In the
long run, psychological theories of therapy must come to a point where
they will make it possible to select the therapy which is good for a pa-
tient and not the patient who is good for a therapy.94 Yet Rogers' suc-
cesses, limited though they may be, clearly show how little we know
about the ego, its activity and passivity, its sources of energy, etc. Reider's
[285] report of "spontaneous cures" likewise shows up our ignorance. In
this sense, though not "embarrassing" or "inconsistent," Rogers seems to
provide data which prod psychoanalysis toward further exploration of
familiar as well as barely charted areas of ego psychology.

Many psychologists and even psychoanalysts (particularly, but not
only, Neo-Freudians) have assumed that Dollard and Miller's [48] study
and Mowrer's [239] experiments and their theoretical combination of
psychoanalysis with learning theory have cut across the "theoretical
jungle" of psychoanalysis, replacing much of it by learning theory. The
powerful position occupied by learning theory until recently (and perhaps
still) on the American scene "reinforced" this assumption. But the fate of
psychoanalytic theory, or for that matter, the fate of any theory, cannot
be settled by popular vote; if it could be, psychoanalysis would be in a
bad way. Learning theory seems to be the (academic) theoretical back-
bone of the majority of recent, mass-produced clinical psychologists. But
since this theory cannot guide their clinical work, they rely there in-
creasingly upon psychoanalytic propositions, whose theory they have not
studied. Thus the "marriage of convenience" that Bollard, Miller, and
Mowrer recommend between psychoanalysis and learning theory must
indeed seem to be convenient to them, since it seems to justify clin-
ical practices, while at the same time it provides a salve to the aca-
demic conscience. Does the work of these authors provide data in-

94 See Knight [202, 203] and Gill [145] concerning the bearing of psycho-
analytic ego psychology on these issues.

The Structure of Psychoanalytic Theory 14!!

consistent with and embarrassing to psychoanalytic theory? Since the]
present themselves primarily as protagonists of psychoanalysis and pur-
port to provide it with a solid experimental and conceptual foundation
this question is not easily answered. Still, we must ask: while these
authors (excepting in some respects Mowrer) do not intend to embarrass
psychoanalysis, have they nevertheless produced data incompatible witib
psychoanalytic findings and theories? Only a brief sketch of the theo-
retical situation can be attempted here [see Rapaport, 271, 2721.

These investigators have produced, by the method of conditioning,
experimental analogues of "Freudian mechanisms" in animals [Masser-
man, 229]. These analogues would be neither embarrassing to nor in-
compatible with psychoanalytic theory7 if no claim were made that in
man, too, the mechanisms of the primary process and of the defenses
are products of conditioning. Bollard, Miller, Mowrer, and Masserman
imply — to say the least — such a claim, and thereby elevate the condition-
ing theory of learning to the status of the learning theory of psycho-
analysis. This is incompatible with psychoanalytic theory, since it makes
the economic and genetic points of view superfluous and thus clashes
with the observational data which made these points of view necessary1
parts of the theory (cf. pp. 110-114). Psychoanalytic theory at present
cannot escape this embarrassment, since it has no learning theory of its
own to pit against conditioning. This lack is not palliated by the demon-
stration that the conditioning theory of learning does not meet the
empirical requirements (e.g., automatization problems, structure forma-
tion, distinction between primary and secondary processes) which a psy-
choanalytic learning theory will have to meet. Psychoanalysis will be
totally free of embarrassment from this quarter only when it has a learn-
ing theory which not only fulfills its own empirical and theoretical re-
quirements, but is also broad enough to account for conditioning
phenomena — including the conditioned analogues of "unconscious me-
chanisms"— as special cases.

The work of these investigators has come up against the problem of
persisting psychic formations, which has beset and embarrassed all
motivational (need-gratification) theories. G. Allport's criticism of
motivational theories and his ego psychology start from this problem, and
psychoanalytic ego psychology faces it squarely. The method of condition-
ing used by Miller, Dollard, etc., determined the form in which they en-
countered this problem: conditioned responses are in general subjed
to extinction ; thus, abiding psychological formations require explanation.
Why are they exempt from this rule? or, How are they so reinforced as
to avoid extinction? This is indeed one of the central difficulties of al
conditioning theories of learning [cf. 82]. The theory of neuroses brings
these questions into sharp relief, since symptoms are apparently non-

146 DAVID RAPAPORT

rewarding and should thus be subject to extinction.95 Bollard and Miller
as well as Mowrer tried to meet the problem by assuming that reinforce-
ment through "learned" (conditioned) drives can account for non-
extinction. This solution brings with it the same difficulties which raised
the problem to begin with — namely, that drives, whether learned or
not, cannot account for persisting structures. But this fact does not seem
to have deterred Bollard and Miller, though they were aware that the
problem is one of ego psychology.

Mowrer, however, was apparently not satisfied with postulating —
and demonstrating by analogues — "acquired drives,'3 but asserted that
they are acquired by contiguity and not by reinforcement learning. This
assertion and the observations it is based on — though questioned by
learning theorists — are a source of embarrassment for psychoanalytic
theory, and will continue to be as long as psychoanalytic theory
accounts for derivative drives by differentiation of basic drives, and infers
that this occurs parallel with structure development, but cannot specify
either the process of structure formation or that of drive differentiation.

But what is more often considered a source of embarrassment for psy-
choanalytic theory in Mowrer's system — his formulation that neurosis is
due to "underlearning" and not to "overlearning" — is actually no source
of inconsistent data. Mowrer apparently saw that a conditioning theory
(whether monistic or dualistic) can hardly explain the persistence of
"learned drives" and nonrewarded symptoms by "overlearning." There-
fore, he reasoned, if the drives and the neurotic drive manifestations can-
not be proved to be "overlearned," then that which is supposed to control
them must be "underlearned." Thus he equated the repressing forces
(censorship, superego) which — according to him — are weak in neuroses,
with underlearned social prohibitions. This sounds logical, but it is not
psychological, and is doubly incompatible with psychoanalytic theory
and observations. First, it implies what is to be proved, namely that the
intrapsychic structures and forces in question are learned (conditioned).
Second, it implies that these structures and forces are ineffective because
of their weakness or absence ("underlearning"), though the concept of
the unconscious in general, and the observations concerning the un-
conscious sense of guilt in particular provide a different explanation so
far not contested by any evidence. Having replaced the "overlearning
theory" which, according to him, is the core of the psychoanalytic "drive-
repression theory" of neurosis, he assumes that he has demolished the
latter. His "underlearning theory" of neurosis, translated into clinical
language, says that the trouble with neurotics is not that their censorship
(repressive forces, conscience) is too strong, but rather that it is too
weak, having been repressed by the id and ego combination. However,

95 N. Maier [228] took this issue so to heart that he propounded a dualistic
theory of motivation learning vs. frustration learning.

The Structure of Psychoanalytic Theory 147

he does not specify the process and mechanism of this "repression." For
Mowrer the neurotic is a person who "did not learn the lesson.35 He takes
us back to moralistic and religious precepts and to the pre-Freudian
conception of neurosis. What seems to have happened is that Mowrer
rediscovered the unconscious sense of guilt, long since recognized in
psychoanalysis. Not realizing the place of his "discovery55 in psycho-
analytic theory, he explained it in terms of learning theory and put it in
the center of psychodynamics, unconcerned with the consequences of
this recentering for psychodynamics at large.

Piaget's observations and theories, if confirmed, may — though they
need not — prove to be sources of incompatible data. They seem to
demonstrate that structure (schema) formation arises from disturbances
in the equilibrium of existing structures (schemata), and that such
disturbances always act as motivations (desirability). In Piaget's terms,
function always brings about structural change (disequilibrium) which,
in turn, provides motivation (desirability) for a repetition of the function
(circular reaction) which consolidates the structural change, i.e., builds
new structure (schema) . Now it may prove possible to treat the observa-
tions on which this theory is based in terms of what psychoanalysis calls
processes of autonomous ego development. If so, Piaget's theory would
shed new light on the nature of many ego motivations and would cor-
roborate Hartmann's assumption that the ego has sources of energy other
than bound and neutralized drive cathexes. It would also force us to
rethink the theory of id-ego relationships. But if Piaget's theory and the
observations it is based on should lead to the conclusion that the only
source of motivations is the one discovered by him, they would become
incompatible with psychoanalytic theory. In either case nothing more
desirable could happen to psychoanalytic theory than a corroboration
of Piaget's findings. Psychoanalysis would find itself for the first time
confronted with a genetic theory of broad scope, using a method of
observation which is in some ways akin to (if not derived from) its own.
The mutual stimulation of this confrontation could not but prove
productive.96

Psychologists, particularly experimental psychologists, seem to assume
that experimental tests of psychoanalytic theories, if negative in outcome,
provide data inconsistent with and embarrassing to the theory. Sears
[308] and the many who have quoted him and relied on him seem to
have assumed something like this. Would that it were so. It is not. Most
of the studies Sears surveyed took a psychoanalytic statement out of its

^Harlow's, Christie's, and others' observations concerning "activity drives"
may also pose a problem akin to that posed by Piaget's studies. But these observa-
tions are too new to be assessed.

148 DAVID RAPAPORT

context and tested the statement, rather than the theory, which they
usually knew little about. Moreover, they used methods of testing alien
to the observations from which the statement and its terms derived. It is
doubtful that any of the currently available experimental results can be
proved clearly incompatible with the theory. Here the very difficulty of
obtaining data inconsistent with and embarrassing to the system becomes
an embarrassment to it. Psychoanalytic theory, which is adequate for
clinical purposes, will have to become much more systematic before ex-
periments can be designed which will not simply confirm or refute its
propositions, but rather specify and modify them. Thus the experimental
psychologist who approaches it must assume the responsibility of clarify-
ing and specifying theoretically the propositions which he undertakes to
test. For the time being this is the only way to arrive at experimental
findings relevant to and incompatible with the theory.

C. "Critical" Tests of Principal Assumptions

Those difficulties in testing psychoanalytic propositions which we
have discussed naturally apply also to the so-called critical tests. In
addition, the latter usually require the existence of alternative theories or
alternative possibilities within the theory. There are few, if any, specific
psychoanalytic propositions for which other theories have an alternative
to offer, and since the psychoanalytic theory itself is not geared to ex-
perimental tests, it does not usually envisage alternatives in the sense im-
plied by the conception of crucial tests, but rather in that implied by
alternative interpretations. While the alternatives in the former sense call
for a decision between two possibilities, one of which is incompatible with
the theory, the alternatives in the latter sense are both consistent with the
theory, but only one is realized in the phenomenon studied, whereas the
other is not. The former pertains to systematic possibilities, the latter
either to a single instance, or to a specific genetic sequence, or to an
individual person. Thus critical tests are hardly possible for the proposi-
tions of the special (clinical) theory of psychoanalysis.97 The opportunities
— if any — for such tests must, then, be sought in the general (psy-
chological) theory of psychoanalysis. But the primitive state of the
systematization of this general theory militates against the possibility of
critical tests, and so does the nonexistence of other theories of com-
parable scope.

For these reasons it is difficult to envisage "critical" tests of this

97 Clinical predictions are always fraught with the fact that all motivations have
multiple, equivalent, alternative means and goals. Thus, such predictions usually
cannot specify which of these equivalent alternatives are to be expected, and
therefore, the results of experimental tests of predictions must first be interpreted
before their bearing on the theory can be established.

The Structure of Psychoanalytic Theory 149

theory. Yet it is probable that the sources of actually or potentially in-
compatible data — i.e., learning theories and Piagefs work — must serve
as points of departure for critical tests.

Critical studies centering around Piaget's theory must first cor-
roborate his observations and extend them to behavior which involves
affects and motivations, in the psychoanalytic sense of these terms. The
aim of studies revolving around learning theory would be to demonstrate
processes of structure building and learning compatible with psycho-
analytic theory, but incompatible with existing learning theories, or vice
versa. Any quantitative method may lead to a critical test if it can trace
qualitatively the process of structure consolidation, that is to say, if it
can show that, once a certain set of qualitative changes has occurred in
a process of acquisition, a nonextinguishing structure arises. It will turn
into a critical test when it can also show that existing learning theories
are incompatible with the process of structure formation traced by it.
None of the well-known methods can at present be regarded as the royal
road to such a critical test. But the following may serve as examples
of techniques which might be tried: tracing quantitatively the qualita-
tive changes in the acquisition of skills which are not simply compounds
of other skills; tracing the qualitative changes in the course of learning
meaningful verbal material; tracing how subjects spontaneously discover
a meaning or a pattern embedded in material wrhich they handle in the
deliberate pursuit of a different goal. Altogether, any quantitative tech-
nique which makes it possible to follow the qualitative (and not just the
quantitative) course of the development of any behavior, which is on its
way to becoming a part of the person's quasi-abiding behavior equip-
ment, might conceivably become the method of choice for a critical test.

This lengthy discussion of "critical" tests is warranted neither by the
actual state of psychoanalytic theory nor by my knowledge of these
matters. Its purpose is to stress that crucial tests — if they are to come —
will not necessarily center on motivations. Indeed, my intention here
has been to make it plausible that the crucial experimental contribution
toward the consolidation of psychoanalytic theory may well be made at
an apparent distance from what is commonly considered its home ground.
It may well come on the battlefield of learning theory, or on that of
perception.

X. METHODS, CONCEPTS, AND PRINCIPLES
OF BROAD APPLICATION

A. The Range of Application

Unlike most psychological theories, whose application outside their
initial ground is a matter of future possibility or probability, the applica-

150 DAVID RAPAPORT

tion of psychoanalysis to nearly all human endeavors and products has
been envisaged and actualized from the very beginning.

The questions here are not what applications are possible, but
rather, how valid and effective are they? How can they be made more
appropriate and effective? Since these applications have rarely been
systematic, the need for and the possibility of studies aimed at systematiz-
ing the existing applications are practically unlimited. Such studies might
well increase the effectiveness of the applications by bringing them in
line with the present state of development and systematization of psy-
choanalytic theory. The development of ego psychology provided psycho-
analysis with new tools which bid fair to increase the appropriateness of
its application in all fields. Particularly the application of psychoanalysis
to sociology* and anthropology has gained and stands to gain further
from Hartmann's [159] and Erikson's [60, 61, 66] contributions, like-
wise, its applications to art from the contributions of Kris [207]. The
change that ego psychology has wrought in the relationship between psy-
choanalysis and psychology has been discussed above.

B. Methods., Concepts, and Principles of Long-term Significance

In his outline, Dr. Koch defines "long-term significance" as the
ability to survive independently from "the over-all structure or detailed
assumptional content of the system." Freud repeatedly stated that an)
therapy which takes into account the unconscious, transference, and
resistances is psychoanalysis. Thus the concepts which he considered tc
be of broadest significance are the dynamic unconscious, transference
and resistance.

But perhaps we can go beyond Freud's view if we consider first
that the methods, principles, and concepts of greatest independence ii
any system are on the one hand those closest to observations and, on th
other, those of greatest generality; second, that some of the methods-
principles, and concepts of all major theories sooner or later become s<
general that they enter the public domain and can no longer be COD
sidered specific to the theory. Psychoanalysis has developed method:
concepts, and principles which are now in the public domain: for e?
ample, the method of interview;98 the concepts of the "descriptive ur
conscious/' motivation, and defense; and the principle of psychologic;
determinism. It could be justly argued that all of these antedate ps^
choanalysis. But psychoanalysis changed their character and gained fc
them an acceptance in the public domain, where they are now ii
dependent of the theory and not subject to its changes.

method of interview implies only that the past is relevant to the unde
standing of the present, but no other specifically psychoanalytic assumption
concept [cf. 166].

The Structure of Psychoanalytic Theory 151

1. Methods. What are the methods of psychoanalysis? It may be use-
ful to distinguish here between methods and techniques. Let us define
techniques as the specific ways and means by which methods are
applied and note that in psychoanalysis they have the additional con-
notation of ways and means which are not only exploratory, but also
effective therapeutically. The techniques of psychoanalysis have been
studied [106, 72, 151] but its methods have scarcely been given system-
atic thought [see, however, Bemfeld, 14]. Thus, what follows can be no
more than a preliminary7 sketch.

It would seem that the basic method of psychoanalysis" is the
method of interpersonal relation [314, 260]; more specifically, it is the
participant observation variant of the method of interpersonal relation;
in particular, it applies the nondirective (free association), the interpre-
tive-genetic, and the defense-analysis techniques of participant observa-
tion [see Gross, 153; also Rapaport, 263, 273]. These methods and tech-
niques, unlike the interview method, are linked to the theory of psycho-
analysis in that the phenomena they are based on are the observational
referents of the transference concept. Human beings in dealing with each
other repeat the patterns they have developed in their relations to
"significant others," and these patterns of relationships ultimately go
back to those which the individual has developed toward the earliest
"significant others" : father, mother, siblings, nurses, etc. Such repetitions
of relationship patterns are the empirical referents of the transference
concept. Transferences are ubiquitous in everyday life, but so far the psy-
choanalytic methods are the only ones for observing them systematically
and for tracing their genetic roots. The aim of the psychoanalytic method
of interpersonal relation is to bring about such transferences. The aim
of the method of participant observation is to make these transferred
patterns conscious. The free association, interpretive-genetic, and defense-
analysis techniques are specific interventions facilitating insight into these
transferences.

In so far as these methods and techniques are tied to the concept of
transference they are specifically psychoanalytic.100 But they are so closely
related to a broad and crucial range of observations that it is hard to
conceive of changes in the structure of psychoanalytic theory which would
alter them or dispense with them. What has changed repeatedly, and is
likely to change again, is the relative emphasis in the theory and in
practice on any one of these methods and on the patient's gaining in-
sight into his transference patterns. Interview and therapy methods which

99 Here we are concerned only with the methods specific to psychoanalysis and
disregard others like suggestion, support, etc. [cf. Bibring, 28].

100 They vary in this respect; of the three, the free-association technique seems
to be the one least closely tied to the concept of transference.

152 DAVID RAPAPORT

do not aim at gaining information about and insight into transference
patterns may well achieve their limited or different goals, but none so
far has succeeded in replacing the psychoanalytic methods of system-
atically observing transference patterns. Projective techniques do obtain
some such data, but the recent emphasis on the significance of the inter-
personal relation between patient and tester [see Schafer, 298] points to
their limitations. Whatever the fate of those more specific methods
described as "techniques," and whatever the ultimate judgment on the
therapeutic effectiveness of these basic psychoanalytic methods, the latter
are likely to stay with us as unique methods of observation for a very
long while.

2. Principles. The "points of view" seem to be the equivalents of
"principles" in psychoanalytic theory. Yet their form shows that the
time to examine them one by one, for their long-range significance, has
not yet arrived. Instead of formal principles we will present here a fewT
general conceptions, which compound the various points of view, and
which seem likely to survive whatever the fate of the more specific
ingredients of the psychoanalytic theory should prove to be.

a. Human behavior is neither merely learned (imprinted by repeated
experience), nor preformed and merely unfolded in the course of a
"maturation" process.

b. Human behavior develops according to the "ground plan"
(Erikson) of an epigenetic process (of which libido development and
ego development are specific aspects) through a sequence of develop-
mental crises, whose solution depends as much on the solutions of
previous crises101 as on the environmental (social) provisions which meet
it (Freud, Hartmann, Erikson, Kardiner, Sullivan).

c. The laws of epigenesis, whose expression in the full perspective
of the individual life cycle is the epigenetic "ground plan," find their
shorter-range expressions in the regulation of all behavior and experience
by intrapsychic motivations and structures. The crucial regulations are
unconscious.

d. The regulation of behavior and experience by motivations and
structures implies : (1) basic tensions (motivations) within the organism,
which strive toward reduction and organize experience and behavior
to that end; (2) basic structures, given by evolution, which on the one
hand serve as guarantees of the organism's adaptedness and adaptation
to the environment (Hartmann, Erikson), and on the other serve as the
means of maintaining, increasing, and discharging the tension which
exists in the organism; they organize experience and behavior to these
ends; (3) differentiation both of the tensions (motivations) and of the

101 Not success or failure but the kind of solution reached is crucial here (Hart-
mann, Erikson).

The Structure of Psychoanalytic Theory 153

structures, in such a manner that the matrix of the differentiation sur-
vives side by side with its products, though its manifestations are always
amplified by these differentiation products; •/4) this differentiation is
determined both by the epigenetic laws and by the environmental (social)
provisions designed to meet the epigenetic crisis in which the differentia-
tion in question comes about; the differentiation products become
further guarantees of the organism's adaptedness and adaptation to the
environment.

It is not implied that no other psychoanalytic propositions have this
degree of generality, nor that other psychoanalytic propositions of equal
or lesser generality may not also have a long-range survival potential.
These four propositions summarize that cohesive core of the most general
conceptions of psychoanalysis (stripped of their specific content) which
has remained constant throughout the changes of the theory and bids
fair to continue to do so. It could be argued that these points are shared
with other psychologies and are not specific to psychoanalysis. This
argument does not hold, though it is a clear indication that psycho-
analytic conceptions have been gradually assimilated by psychology at
large. No other psychology contains this assembly of general conceptions,
methods, concepts, and theories; nor has any other psychology supported
any of these by as broad an array of observations as has psychoanalysis.

3. Concepts. The major concepts of a high survival potential per-
taining to each of the metapsychological points of view are :

a. Dynamic point of view. The concepts of unconscious forces and
conflicts are close to observations and yet of sufficient generality to have
a high survival potential. The concepts of drive, drive-fusion, specific
drives (sex, aggression, life and death instincts, etc.) are of a lesser gen-
erality, and may well change or be replaced as the theory changes.

b. Economic point of view. The concepts of primary process, second-
ary process, and pleasure principle (wish-fulfillment) are so directly
related to observation and so general that they are likely to survive. The
concepts of cathexis, binding, and neutralization, however, are both more
inferential and more specific, and while the observations do seem to
demand some set of quasi-quantitative concepts like these, it is uncertain
whether they will survive in their present form.

c. Structural point of view. The concepts of structure and relative
autonomy (Hartmann) are indispensable to the theory, and at present
it is not possible to foresee changes in the theory which could eliminate
them. But the concepts of id, ego, superego, and the differentiation of
the ego into defense-, control-, and means-structures are neither as in-
dispensable to nor as independent from the theory. However, a variety
of subordinate structural concepts (e.g., specific primary-process and
defense mechanisms, like displacement, condensation, substitution, sym-

154

DAVID RAPAPORT

bolization, repression, isolation, reaction formation, projection)102 which
are more directly related to observations and of a lesser generality, are
likely to survive. It is not implied, however, that this holds for all the
specific defense mechanisms.

d. Genetic point of view. We discussed above the high survival
potential of the epigenetic principle. This holds also for the conception of
the crucial role of early experiences, as well as for the concepts of fixation
and regression. The specific concepts related to libido development, such
as orality and anality, are also likely to survive [cf. Kardiner, 194, 195;
Sullivan, 313; Erikson, 56, 61, 62], since they are closely related to
observations. However, the classic conception of libido development itself
may well undergo radical change, as it becomes one aspect of the integral
process of epigenesis. The conception of the special role of psycho-
sexuality, even though it has good empirical anchorage, does not seem to
have that degree of generality which would make it a theoretical neces-
sity (cf. pp. 89-90, above).

e. Adaptive point of view. The conceptions of the organism's pre-
paredness for an average expectable environment (Hartmann), ap-
paratuses of primary and secondary autonomy (Hartmann), mutuality
(Erikson), relative autonomy from the environment (Gill-Rapaport),
i.e., the dependence of the secondary process on external stimulation [98,
p, 515], modes and modalities (Erikson), though too new to be properly
evaluated, do seem likely to survive.

Now, in brief, about the concepts of the special (clinical) theory. Let
us take the transference concept as an example. We have encountered
it as the foundation of the long-range significance of psychoanalytic
methods. Yet its own survival potential might be characterized as bor-
rowed. The referent of the transference concept is not a single process but
a congeries of processes. The patterns which are transferred may be broad
or fragmentary, and the processes by which transference is accomplished
are many and varied: wish-fulfillments, displacements, projections, etc.
The transference concept refers to an end result: it is an achievement
concept. In the clinical theory of psychoanalysis it is indispensable, but
the general theory of psychoanalysis resolves it into process concepts. It
is probably not far off the mark to suggest that this is the case for most
clinical concepts. A case in point is the very definition of resistance as
the manifestation of defense. It is not implied, however, that clinical
concepts do not have a survival potential : they do, but only when they
are close to observations and when the process-concepts which underlie
them are themselves likely to survive. A study of the concepts of the
clinical theory from this point of view would be rewarding, but so far we

^Objections might be raised against discussing these primary-process mech-
anisms as structures, but I cannot attempt to justify this here.

The Structure of Psychoanalytic Theory 155

do not have even a workable separation of the special clinical and the
general psychological concepts of psychoanalysis.

XL THE THEORY'S ACHIEVEMENTS AND ITS CONVERGENCE
WITH OTHER THEORIES

A. Achievements

Freud's earliest program [94, appendix] was to develop a general
psychology on neuropsychological lines. This attempt failed and Freud
concluded [94, 64] that the theory of behavior must be a psychological
theory. But he never gave up the belief that once psychoanalysis had
developed far enough, its link to physico-chemico-biological processes
would be found. Apparently the time for this has not yet arrived and
the recurrent popularity of neurological models has so far not brought
it any closer. It seems that until psychology has progressed much further,
attempts at neurological or biological explanations of behavior are
bound to be of little avail. Freud's program to develop a general psy-
chology receded into the background for a while, but was revived with
the development of ego psychology.

To solve the problem of conversion (somatic compliance) was also
part of the earliest program. The nature of the hysterical conversion
symptom — the psychological conflict's "leap into the somatic" — was and
has remained a haunting riddle, though Freud began early to question
that psychoanalytic methods and theory could solve it. Psychoanalysts,
instead of solving the problem, generalized it, first into the conception of
"organ neurosis" [F. Deutsch, 45; Meng, 230], and then into "psycho-
somatic medicine" [Alexander and French, 4, 5, 86, 90; Dunbar, 49, 50;
Weiss and English, 321]. The number of investigators and investigations
in this field is great, broad areas of observation have been scouted and
mapped, and the effect on medicine proper is considerable, but it is not
clear just how much — if any — theoretical advance has been made. Psy-
chosomatic studies remain fraught with the problem of "specificity,"
which so far has defied solution. It is worth noting, however, that rela-
tively recently a possible clue to the conversion riddle has appeared
[see Travell, 317].

The programs so far discussed belong to that phase of Freud's work
which was preparatory to psychoanalysis. The main program Freud set
for psychoanalysis proper (1900) was to explore the unconscious; later
(1923) this changed into the exploration of the id and the unconscious
ego. Discoveries are still being made in both areas and much of the "un-
conscious ego" is still uncharted territory. Yet considering that successful
exploration always breeds new problems, the work on this program

156 DAVID RAPAPORT

can be considered well advanced. This, however, is a judgment within
the frame of reference of psychoanalysis: it refers to the program of ex-
ploration and not to a program of testing and developing the theory by
means of experiments.

A related program was to apply the theory to myth, legend, literature,
art, ethnology, etc., in order to demonstrate its pertinence to all human
behavior and products and thus to obtain a broad base of supporting
evidence. The achievements in this direction have already been men-
tioned, and the importance of the new means provided by ego psy-
chology for the further pursuit of this program has been indicated.

For a long while the exploration of the ego seemed to be only a
contemplated program. Freud expected the information about ego
functions to come from the study of "narcissistic neuroses" (i.e., psy-
choses), but delayed this study because he considered the exploration of
the unconscious to be the primary7 task. Yet this program was indirectly-
pursued throughout the history of psychoanalysis in the study of the
defenses, censorship, secondary process, and reality relationships. How-
ever, Freud did embark (1921, 1923) on an explicit conceptualization
of the ego without studying psychoses anew, apparently prompted by the
problem of "the negative therapeutic reaction53 and "the unconscious
sense of guilt" [124, 126]. Later he carried the study of the ego further
(1926) by re-evaluating the problem of anxiety [131]. Other psycho-
analysts followed his lead [77, 245, 284, 320] and the achievements of
this phase of the program were capped (1936) by A. Freud's [93] work.
The ego-psychological program was then dramatically broadened by
Hartmann [157] and by Erikson [56, 59]. The ego was explored slowly
but so successfully that a broad and still uncharted area was opened up.

The program of superego exploration was already implicit in the
study of censorship (1900). But only the study of narcissism [114]
brought it into focus in the term "ego ideal" (1914). Though the con-
cept of the superego was formalized simultaneously with the ego [124,
126], and in spite of significant advances [see 73], the work on this pro-
gram has hardly passed the beginning stages.

While psychoanalysis as a therapy is primarily the subject matter
of the special (clinical) theory, the theory of therapeutic technique is
part of the general theoretical program of psychoanalysis. It was so
treated by Freud in the prehistory of psychoanalysis [35, chap. 3], in
some of the "Papers on Technique" [106] and in "Analysis Terminable
and Interminable" [139]. Nevertheless, this program is still far from ful-
fillment. Even the most systematic [72] of the few extensive [151, 224]
treatments of technique contributes little toward the theoretical pro-
gram. E. Bibring [28] has penetrated into these problems further than
most others. Recently Eissler [51] and Gill [145] have also made rele-

The Structure of Psychoanalytic Theory 157

vant contributions. The importance of this program and the difficulties
in its way are equally great. Progress may require not only the study of
the techniques of psychoanalysis and those of other schools of therapy,
but also the development of a psychoanalytic theory of communication.

The theoretical explanation of neuroses was an outstanding part of
the program. This is where the work on the special (clinical) theory of
psychoanalysis had its greatest achievements and also brought a con-
siderable general theoretical yield, to which FenichePs [735 77] systematic
survey of the special theory refers continuously. Yet we still do not have
a systematic treatment of neuroses, from the point of view of the general
theory, comparable to that which Freud gave of dreams in chapter 7
of The Interpretation of Dreams [98].

The theoretical explanation of psychoses was also a part of the pro-
gram. Beginning with his early (1895) study of a case of paranoia [941
and his analysis of the Schreber case [107], Freud dwelt on it repeatedly
[114, 117, 119, 126, 127, 128]. Yet despite the contributions of Abraham
[1], Federn [70], B. Lewin [2141, Fromm-Reichmann [142, 143],
M. Wexler [324, 325], Hartmann [163], and others, and in spite of the
studies by Putnam, Mahler, Bettelheim, and other psychoanalysts on
juvenile schizophrenia, the fulfillment of this program has barely begun.

The situation is only slightly better in that part of the program which
comprises the general theory7 of character disorders, addictions, delin-
quency, criminality, and borderline problems [22, 23, 202, 203, 282,
283]. J

B. Convergence with Other Theories

It is difficult to differentiate the applications of the theory to other
fields, its influence on other sciences, and its convergence with other
theories and sciences. The distinction might be drawn, perhaps, as
follows: application is the work of psychoanalysts in other fields; in-
fluence is the adoption of psychoanalytic assumptions, methods, findings,
and/or theories by workers in other fields; convergence is mutual in-
fluence.

In this sense, in anthropology the days of application [Freud, 109;
Roheim, 289, 291, etc.] and influence (e.g., Kluckhohn) are past, and
convergence can be observed on one side in Erikson's work, and on the
other in the work of cultural anthropologists (however the opinions may
vary about this work otherwise). [See also Psychoanalysis and the Social
Sciences, 290.]

The same holds for sociology, where the days of application and in-
fluence (Freud, W. Reich, Fenichel, Lasswell, and the early Fromm)
are past and convergence can be observed, for instance, in Parsons',
Riesman's, and N. Foote's work on the one hand (however the opinions

158 DAVID RAPAPORT

may vary about this work otherwise), and in Hartmann's and Erikson's
on the other.

The convergence of psychoanalysis with medicine in general and
psychiatry in particular, though only too obvious, is practical rather than
theoretical.

It is questionable whether one can speak of a convergence of psycho-
analysis with the other fields, its applications to which wrere mentioned
earlier: in art, literature, history, etc., we find influence, but no more.

Now, to the convergence of psychoanalysis with psychology. We have
already mentioned that psychoanalytic ego psychology seems to be re-
sponsible for a considerable part of this. The convergence with develop-
mental psychology is of long standing (Werner and the early Piaget)
and is reinforced on the one hand by the recent longitudinal and cross-
sectional studies of psychoanalysts like M. Fries, Spitz, Escalona and
Leitch, Benjamin, Kris, Mittelmann; and on the other by the studies of
Piaget, Werner and his associates, and others. The work of K. Lewin and
his associates on the one hand, and that of T. French on the other are
outstanding indications of convergence. The studies in learning theory
by Bollard, Miller, Mowrer, Sears, etc. (however the opinions may vary
about this work otherwise), represent a convergence of psychology with
psychoanalysis. Murray's early work, his and his associates' and their
successors' work in the assessment of personality are also indications of
this convergence; so is much of the recent work in experimental clinical
psychology. Two other important indications of convergence, the studies
on motivations and memory and on motivations and perception, have
already been discussed.

The future of this convergence may hinge on the solution of the
problem of structure formation and learning. If that solution should
arise from the matrix of psychoanalytic theory, the latter may become
the core of psychology proper. If the solution should prove relatively
independent of psychoanalysis, then the latter is likely to become a rela-
tively subordinate part of the general theory of psychology as the core
of its clinical and motivational theories, but its concepts and theories will
be reducible to more fundamental ones. The existing learning theories
have not accomplished this reduction and it seems unlikely that they ever
will.

Finally, coming closest to the home base of psychoanalysis, the devel-
opment of psychoanalytic ego psychology has begun to extract the valid
contributions from the theories of the Neo-Freudian schools, and thus
to initiate the convergence of these offshoots with psychoanalytic theory
proper. There is still much to be done here and the convergence pertains
only to the theories, not to the "schools" as organizations of vested
interest.

The Structure of Psychoanalytic Theory 159

XII. TASKS FOR THE FUTURE DEVELOPMENT

OF THE THEORY

A. Empirical Evidence Needed

Psychoanalytic theory does not need additional data per se for its de-
velopment: the amount of data is already embarrassingly large. It does
not need the Blacky Test type of data which, though amenable to
statistical treatment, are simply masked clinical data: the clinical data
are better. It does not need experimental data which replicate clinical
relationships. What it needs are methods to obtain data which can lead
beyond the clinical relationships to theoretical relationships of the type
discussed in this essay.

It is for this reason that experimental data on structure formation
and learning, and data corroborating or negating Piaget's observations
and theories are needed.

In Hartmann's theory, we have for the first time a concept of auton-
omy and we have (cf. pp. 95ff. and p. 98, above) also an elaboration
of it into a threefold conception of relative autonomy (Erikson, Gill).
The methods used in the McGill University and Bethesda studies on
sensory deprivation, as well as the hypnotic methods, seem to be ap-
propriate means to alter the balance of these relative autonomies [see
280]. The theory needs data obtained by these or other relevant methods
(e.g., drugs like mescaline), but just any data obtainable by these
methods will not do: the need is for data obtained in controlled experi-
ments guided by the theory of autonomy.

In Hartmann's studies we have for the first time a concept of conflict-
free ego functions. G. S. Klein's work contains a variety of methods for
the study of those ego structures which, unlike defenses, are conflict free
and serve to control and channel motivations. Data concerning such
structures are needed. But an indiscriminate proliferation of such data
would provide mainly a catalogue of "cognitive attitudes" (Klein), just
as French psychiatry has provided us with a term for more or less every
possible form of phobia. The data needed are those which will elucidate
the relation of these style structures to other ego structures (e.g., de-
fenses), to motivations, and to each other.

In Erikson's work we have for the first time a theory and an epi-
genetic ground plan of ego development. Additional data concerning
each epigenetic phase are needed. But again, just any data pertaining to
a given phase of development will not do. The data should pertain to
Erikson's observations and should corroborate, elaborate, modify, or
negate them. To obtain such data, the investigators will have to adopt
Erikson's frame of reference, at least to begin with.

We suggested above that a prerequisite of the theory of therapeutic

160 DAVID RAPAPORT

technique may be a theory of communication. Data for building such a
theory are needed. The data and concepts of the existing attempts at a
communication theory do not seem to be relevant. The focus of such a
communication theory7 must be the laws which govern the tendency of
communication to engender or to prevent reciprocal communication.
Moreover, it should be a theory7 in which the communicants' becoming
conscious of something is equivalent to (latent) verbal or nonverbal
communication [see 263, 273]. The methods by which data relevant
to such a theory can be obtained have yet to be worked out.

In Hartmann's "self [161, 162]' and in Erikson's "identity" [66]
we have in psychoanalytic theory for the first time concepts to account
for the historical continuity of the individual and for his self-experience,
and conceptual tools to distinguish them from the referents of the ego
concept. Data pertaining to and permitting the elaboration of these con-
cepts are needed. But just any data of "self-experience," "self-evalua-
tion," or "ego-involvement" will not do. They must be data concerning
the relation of the "self5 or of "identity" to the psychoanalytic theory of
psychological functions in general and of ego functions in particular.

The less than satisfactory progress in the theoretical understanding
of schizophrenia and other psychoses has been mentioned. Here again
data are in abundance. What to do with them is the question. They have
not been selected to reveal the relation of the phenomena of schizo-
phrenia to the existing theory. There is no need for more data showing
that the content of psychotic products can be interpreted like dreams or
unconscious fantasies. Nor are data needed on oral or anal wishes under-
lying the manifest content of psychotic products; these are ubiquitous
in man, and only their role, intensity, and frequency might conceivably
be specific to a given psychosis. It is the formal characteristics of psy-
chotic behavior (action, affect, and thought) which seem to be specific,
and what is needed are data to connect them with the psychoanalytic
theory.

Data are needed to reveal the similarities and differences between
analogous structures (and motivations) on different hierarchic levels of
the psychological organization. J. F. Brown [36] obtained some data of
this kind [cf. also 276].

Last but not least, though no data replicating clinical relationships
are needed, any replication whose purpose is to quantify these relation-
ships so as to pave the way toward "dimensional quantification" should
be welcome.

This enumeration has no systematic pretensions, nor does its sequence
imply an order of importance. The examples were chosen to show that,
more than data, we need methods which promise to yield data relevant
to the theory and its unsolved problems.

The Structure of Psychoanalytic Theory 161

B. Obstacles to the Development of the Theory

Here we will dwell on theoretical obstacles, leaving the discussion of
practical obstacles for Section C.

The days of the resigned belief that complex psychological phenom-
ena cannot be studied in the laboratory are past. So is the overenthu-
siasm of K. Lewin's early days [220], when there seemed to be no doubt
that all psychological phenomena could be relevantly studied on the
laboratory scale. While we all hope that even- psychological phenom-
enon Is amenable to scientific study, to find the ways and means for this
has become our gravest concern. The mam obstacles to the development
of psychoanalytic theory center around these ways and means.

First y due regard for the individual's rights sets limits to the manipu-
lation of beha\ior outside and even inside the laboratory; and due regard
for the privacy of the individual sets limits even to observation. This Is
one of the major empirical obstacles. The problem Is not only the ethical
one of trespassing on rights and privacy but also, and perhaps primarily,
what such trespassing does to the subject, to the observer, and to the
observation.

Second, the hierarchic problem, so heavily stressed In these pages,
implies that reduction to laboratory size more often than not changes
the hierarchic position of the phenomenon or relationship in question,
so that not the phenomenon or relationship itself, but a high-level hier-
archic equivalent of it is studied. This is not simply an obstacle. It indi-
cates that laboratory research can attack all psychological problems,
provided it centers its attention on the laws of hierarchic transformations.
Once such laws begin to take shape, psychologists will be able to dis-
pense with the arbitrary claim that the laboratory findings obtain for
life situations and will use these laws as the rules by which inferences
from the laboratory findings to life situations can be drawn. This the-
oretical complexity then is not per se an obstacle, though there is long
and arduous experimentation ahead before these laws of hierarchic rela-
tions are discovered and brought to a point where they can serve as rules
of inference.

Third, laboratory methods cannot get around the troublesome fact
that there are many psychological phenomena which occur, as a rule,
only in the contact of one person with another (or others). The study of
such phenomena led to the method of participant observation in therapy,
in everyday life, and in laboratory situations. This method has scarcely
been explored theoretically; in it the investigator enters into the privacy
of the subject, but he does so at the price of becoming a participant,
shouldering all those implicit and explicit commitments which participa-
tion involves. Psychoanalysts and other therapists know a great deal

162 DAVID RAPAPORT

about these commitments and their effect on the observer and on the ob-
served. But the implications of this knowledge for the method have not
yet been theoretically formulated [see, however, Bemfeld, 14, Gross,
153], and the lack of such systematization is an obstacle in the way of
the theory's development.

We have discussed the possibility that experimental study will dis-
cover rules of inference, by means of which conclusions can be drawn
from laboratory-sized to life-sized phenomena. What about the rules of
inference for relating data obtained by direct observation to data ob-
tained from participant observation? For instance, the psychoanalytic
theory of development is built from reconstructions based on data ob-
tained by the method of interpersonal participant observation in the
therapeutic two-group situation, while Piaget's theory of development
is built on data obtained in direct observation.103 Now it is possible that
the theories of development of psychoanalysis and of Piaget will prove
compatible, and rules of inference will be found to link their concepts.
Indeed, it is possible that their mutual influence will lead to a redefini-
tion of their concepts so that there will arise a single conceptual system
which subsumes both theories or subsumes one under the other. But
there are two other possibilities. First, the two theories might prove in-
compatible and thus one of them untenable. Second, it might just hap-
pen that the two theories, like the observations they are based on, will
prove not to overlap, and not to be incompatible. If so, the two methods
will have arrived at theories pertaining to two different aspects of the
same subject matter. We might, then, have to conclude that these two
aspects of the subject matter are complementary [cf. Niels Bohr's com-
plementarity concept in atomic physics, and the complementarity in the
study of the living cell envisaged by the biophysicist Delbrueck, 44].
The uncertainty whether the yield of the participant observation method
and the yield of other methods can be related to each other by conjunc-
tive rules of inference, or must be related by a disjunctive rule of comple-
mentarity, is a major hurdle in the way of the development of psycho-

10SA11 observations on human beings are in a sense participant observations:
one-way screens, movies, and sound tracks obscure but do not circumvent this fact.
Yet there is a difference between being a participant observer and using the
method of interpersonal participant observation, and there is also a difference
between constructing and reconstructing developmental relationships from partici-
pant observations. Piaget [254, 255, 256], too, was a participant observer: by his
actions, he modified the situations and the tasks his children faced. Yet he did
not use the method of participant observation, in that he did not systematically
study the changes in the children's relation to him (their father) consequent to
his "participation/' nor the changes in their sensorimotor behavior as the latter
depended on the children's relation to him.

The Structure of Psychoanalytic Theory 163

analytic theory. It is possible that an inkling of this difficulty accounted
for Freud's lack of interest in the attempts to verify psychoanalytic prop-
ositions by methods other than psychoanalytic.

Fourth, the last of the obstacles to be mentioned here is the problem
of mathematization, including quantification, already discussed. It is
both an empirical and a theoretical obstacle to the development of the
theory.

C. The Practical Obstacles to Theoretical Advance in Psychology

Let us first take the obstacles in the way of psychoanalytic theory.
Here the lack of systematic theoretical literature, the nature of psy-
choanalytic training, and the character of psychoanalytic practice stand
out.

As a rule, the observer and experimenter is guided in his contribu-
tion to theory development by systematic theoretical literature. In the
lack of such, the investigator has to master the primary sources and do
the systematization for himself. This is a time-consuming pursuit to
which the habits of psychoanalytic practice are not conducive. It is often
said that psychoanalytic theory is a rigid and unchangeable doctrine.
Although there is such dogmatism and orthodoxy in the Societies and
Institutes (whether they are Freudian or Neo-Freudian or in between)
in regard to the clinical theory, I have rarely found dogmatism in regard
to the general theory. The attitudes range from enthusiasm, through lack
of interest, to total lack of information. The general theory, far from
being well-ingrained dogma, is a waif unknown to many, noticed by
some, and closely familiar to few. Not the alleged rigidity of the theory,
but rather unfamiliarity with it is the obstacle to theoretical progress.
The lack of systematic theoretical literature is certainly not the sole cause
of this situation (the original sources are available) but it is a major
handicap to advancement.

The training given by psychoanalytic Institutes is primarily designed
for future practitioners, and limited to physicians. The scope of this
training is defined by several factors: (a) its "night school" character,
(b) the average medical training, which prepares the students neither
for psychology, psychiatry, and psychoanalysis, nor for theoretical and
research pursuits, (c) the fact that both teachers and students are, as a
rule, full-time practitioners. Two additional facts about this training:
first, it is postdoctoral, time-consuming, and costly, and thus pushes the
graduate to seek more lucrative and less leisurely pursuits than research;
second, though the rules limit it to physicians, some psychologists and
other scientists can obtain "research training" in psychoanalysis, but this
includes only training analysis and course work, not supervision (control)

164

DAVID RAPAPORT

and often not even clinical seminars,104 although the theoretician and
research man needs full training no less than does the future practitioner.
Thus the "medical closed shop55 works doubly against progress in psycho-
analytic theory. It is small wonder that divergences of observation and
thinking among psychoanalysts tend to be resolved not by theoretical
or empirical decision but by orthodoxy and secessions.

The nature of psychoanalytic practice does not foster theoretical
development. The long workdays, spent closeted with patients, provide
neither the necessary time and leisure nor the detachment. The solitary
character of the practice minimizes that kind of collegial interchange
which is the fertile soil of theory making. The grants available and the
institutions which relieve some psychoanalysts from the burden of full-
time practice and provide opportunities for such interchange are for the
privileged few. While no science has more than a few theoreticians at
a time, those few always emerge from the many who try. Where only a
few can try, the prospects remain dim, however well or poorly the few
may be chosen.

The effect on the psychoanalyst of the limitation on the number of
patients he can see is enhanced by the limited range of people con-
sidered treatable by psychoanalysis and by the limited number who can
afford it. Moreover, the outstanding psychoanalysts sooner or later be-
come training analysts, and then part of their time is occupied with an
even more limited group: the kind of people who want to become psy-
choanalysts and pass through the sieve of the training committees. These
limitations are particularly crippling to the development of the psycho-
social aspects of the theory, but they also leave psychoanalysis centered
on its clinical aspects, to the neglect of its general theory. True, the clin-
ical theory needs further development and its methods are so far indis-
pensable for the study of a wide range of phenomena; but there is
another wide range of phenomena, crucial for the development of the
general theory, that is not amenable to study in the therapeutic situation
where the patient's interest is — and should be — the guide. The develop-
ment of ego psychology is particularly affected by this limitation.

It seems that without more scholarly and academic training, and
without the admission of nonmedical students to such training, the main
obstacles in the way of the development of psychoanalytic theory are
bound to persist. It is unlikely that medical schools or psychology de-
partments would do better than the psychoanalytic Institutes: neither
their traditions, nor their chances of recruiting training staffs, nor the
complexity of the training problem to be met seems to bode well for
such "simple" solutions.

104 Since this was written, initial steps have been taken by the American
Psychoanalytic Association to explore ways to change this situation.

The Structure of Psychoanalytic Theory 165

Now, the obstacles to theoretical progress in psychology: the "sci-
entific method," the addiction to a single method (or limited set of
methods) , and the measuring rage stand out.

Theory making, i.e., theoretical progress, begins in familiarity with
phenomena and in thinking about them for about the theories pertain-
ing to them). It continues in hunches and speculations, some of which
are amenable to empirical test; others, which spin relations between con-
cepts and theories, or restructure and systematize them, are not and need
not be, though they may well lead to conclusions \vhkh again can and
must be subjected to empirical test.

The "scientific method" is the canon by which that record is made
which we call science the codified, interconnected body of accepted
knowledge. But it is not the canon for making discoveries, nor the canon
for making theories. Nor is the canon, by which the scientific record is
made, unique and static: it changes with the change in the methods,
subject matter, and aims of research. Dingle, the British historian of
science, had harsher words about the "scientific method, or methodol-
ogy as it is often called now" :

... a discipline conducted for the most part by logicians unacquainted
with the practice of science, and it consists mainly of a set of principles by
which accepted conclusions can best be reached by those who already know
them. When we compare these principles with the steps by which the dis-
coveries were actually made we find scarcely a single instance in which there
is the slightest resemblance. If experience is to be any guide to us at all — and
what scientist can think otherwise — we must conclude that there is only one
scientific method: produce a genius and let him do what he likes . . . the
best we can do is to learn to spot natural genius . . . and protect it, by
fiery dragons if need be, from the god of planning [46, pp. 38-39].

Beveridge [24] described scientific investigation as an art. Theory
making may be described as a work of imagination; the "scientific
method" comes into play only in testing the theory and in making the
record. But even there, however much the scientific method can help to
design economic and valid tests, the essential ingredient is still the in-
genuity in inventing a method which connects the phenomena and the
theory.

The stress on the "scientific method" becomes an obstacle to the-
oretical advance in several ways. First, the stress on teaching the scien-
tific method and the design of experiment diverts attention from training
in observation. Second, it discourages the budding investigator's interest
and trust in his own hunches and speculations. Third, it makes the "sci-
entific method" and the "design of experiment" appear as a sure-fire
way to produce "research findings." The findings thus produced clutter
our literature and crowd out the interest in methods of experimenting

166 DAVID RAPAPORT

and observing. Fourth, it leads to a publication policy (and, through it,
to a training by precept) such that the publications conform to the "sci-
entific method55 and cover up the actual tracks of the investigator even
when by chance his tracks would be worth knowing. The publications
read as though investigation consists of nothing but the application of
the scientific method. Thus to the novice, our (and what is more im-
portant, his own) actual disorderly ways of productive thinking appear
as an inadequacy. His self-observations, which show him that his think-
ing does not follow the "scientific method," become the sources of a
gnawing self-doubt, which in turn only too often leads to a sterilizing
discipline of thought. No wonder that in our literature few authors are
surprised, few things are surprising, and a deadly boredom prevails, aided
and abetted by what the given journal considers to be the form of sci-
entific reporting.

The bane of the "single theory and single method" is in part synon-
ymous with the plague called "schools of psychology." The investigator
uses a method and becomes its captive. So do his students. He develops
a theory which can only predict phenomena elicited by that method or
a closely related one. What is not amenable to study by those methods
ceases to influence the theory. In turn, all theories whose methods do not
apply to the realm of phenomena in question are somehow considered
"wrong," and if they are tested at all, it is by methods alien to them,
and so they are obviously found wrong. Usually, however, they are ig-
nored altogether. As a result, certain methods become "canonized," the
study of a limited range of phenomena becomes the only "proper study
of man," and those who try to reunite the field of psychology, so frag-
mented by a few methods, are regarded as "philosophers" in the pejora-
tive sense of the word. To be a theorist becomes an opprobrium : this is
the particular form of anti-intellectualism which is endemic in present-
day psychology. No new methods (i.e., ways of experimenting, in contra-
distinction to designs of experiment) are sought to break the splendid
isolation of the self-encapsulated realms of phenomena thus created.
Methodological thinking, which deals with the relation of method and
theory, and attempts to establish what is an artifact of the investigative
method and what is "the nature of the beast," remains mostly beyond
the ken of the psychologist.105

The "measuring rage," already discussed, is particularly character-
istic of the experimental work in clinical and personality psychology. It
expresses and fosters a disregard for theory, and is thus a major obstacle

300 This methodological implication of Brunswik's "representative sampling of
design" is often overlooked. For brevity and emphasis, I deliberately overstate
these points: for instance, the "artifact" issue is by no means as simple as the
above statement suggests.

The Structure of Psychoanalytic Theory 167

to theoretical advancement. But It also distracts attention from the gen-
eral problem of mathernatlzation and the specific problem of dimensional
quantification. We may not be too far off the mark In suggesting that the
malaise of psychology which is manifested In the "measuring rage" is the
same as the one responsible for the epidemic-like popularity in psychol-
ogy of ''information theory," "open systems/5 "stress syndrome," and
other extrapsychological achievements. Conceptions and methods can
be borrowed from other sciences: all that Is useful should be used. But
the epidemic of grasping at every likely new achievement of other sci-
ences seems to be a symptomatic giveaway: salvation is expected from
the outside and not from results achieved by the sweat of our own
brows. At the root of It Is a lack of self-confidence : the lack of assurance
that psychology knows where It has come from and where It is going.

REFERENCES

1. Abraham, K. Selected papers of . . . . London: Hogarth. 1927.

2. Adams, D. K. A restatement of the problem of learning. Brit. J.
Psychol., 1931, 22, 150-178.

3. Alexander, F. The relation of structural and instinctual conflicts.
Psychoanal Quart. ,1933, 2, 181-207.

4. Alexander, F. The influence of psychologic factors upon gastro-
intestinal disturbances: a symposium. 1. General principles, objectives, and
preliminary results. Psychoanal. Quart., 1934, 3, 501-539.

5. Alexander, F. Emotional factors in essential hypertension, presenta-
tion of a tentative hypothesis. Psychosom. Med., 1939, 1, 173-179.

6. Allport, G. W. Personality. New York: Holt, 1937.

7. Allport, G. W. Scientific models and human morals. Psychol. Rev.,
1947, 54, 182-192. Also in [8, pp. 187-197].

8. Allport, G. W, The nature of personality: selected papers. Reading,
Mass.: Addison- Wesley, 1950.

9. Allport, G. W. Becoming: basic considerations for a psychology of
personality. New Haven3 Conn.: Yale Univer. Press, 1955.

10. Beach, F. A., & Jaynes, J. Effects of early experience upon the be-
havior of animals. Psychol Bull, 1954, 51, 239-263.

11. Benjamin, J. D. Methodological considerations in the validation and
elaboration of psychoanalytical personality theory. Amer. J. Orthopsychiat.,
1950, 20, 139-156.

12. Bergman, P. A religious conversion in the course of psychotherapy.
Amer. J. Psychother., 1953, 7, 41-58.

13. Bernfeld, S. Die Gestalttheorie. Imago, 1934, 20, 32-77.

14. Bernfeld, S. The facts of observation in psychoanalysis. /. Psychol.,
1941, 12, 289-305.

15. Bernfeld, S. Freud's earliest theories and the school of Helmholtz.
Psychoanal Quart., 1941, 13, 341-362.

168 DAVID RAPAPORT

16. Bernfeld, S. Freud's scientific beginnings. Amer. Imago, 1949, 6,
163-196.

17. Bernfeld, S. Sigmund Freud, M.D., 1882-1885. Int. J. Psycho-Anal,
1951, 32, 204-217.

18. Bernfeld, S., & Bemfeld, S. G. Freud's first year in practice, 1886-
1887. Bull Menninger Clin., 1952, 16, 37-49.

19. Bemheim, H. Die Suggestion und ihre Heilwirkung. S. Freud
(Trans.). (2ded.) Leipzig: Deuticke, 1896.

20. Bertalanffy, L. von. The theory of open systems in physics and
biology. Science, 1950, 111, 23-30.

21. Bettelheim, B. Individual and mass behavior in extreme situations.
/. abnorm. soc. Psychol, 1943, 38, 417-452.

22. Bettelheim, B. Love is not enough: the treatment of emotionally
disturbed children. Glencoe, 111.: Free Press, 1950.

23. Bettelheim, B. Truants from life: the rehabilitation of emotionally
disturbed children. Glencoe, 111.: Free Press, 1955.

24. Beveridge, W. I. B. The art of scientific investigation. New York:
Norton, 1950.

25. Bexton, W. H., Heron, W., & Scott, T. H. Effects of decreased
variation in the sensory environment. Canad. J. PsychoL, 1954, 8, 70-76.

26. Bibring, E. The development and problems of the theory of the
instincts. Int. /. Psycho-Anal, 1941, 22, 102-131.

27. Bibring, E. The mechanism of depression. In P. Greenacre (Ed.),
Affective disorders. New York: International Universities Press, 1953. Pp.
13-48.

28. Bibring, E. Psychoanalysis and the dynamic psychotherapies. /.
Amer. psychoanal Ass., 1954, 2, 745-770.

29. Blake, R., & Ramsey, G. (Eds.) Perception: an approach to per-
sonality. New York: Ronald, 1951.

30. Bond, D. D. The love and fear of flying. New York: International
Universities Press, 1952.

31. Boring, E. G. A history of experimental psychology. (2d ed.) New
York: Appleton-Century-Crofts, 1950.

32. Boring, E. G. Dual role of the Zeitgeist in scientific creativity. Sci.
Monthly, 1955, 80, 101-106.

33. Brenman, M., & Gill, M. M. Hypnotherapy. New York: Interna-
tional Universities Press, 1947.

34. Brenman, M., & Gill, M. M. Research in psychotherapy round table,
1947. Amer. ]. Orthopsychiat., 1948, 18, 92-118.

35. Breuer, J., & Freud, S. Studies in hysteria. New York: Nervous and
Mental Disease Publications, 1937.

36. Brown, J. F. Ueber die dynamischen Eigenschaften der Realitaets-
und Irrealitaetsschichten. Psychol Forsch., 1933, 18, 2-26.

37. Bruner, J., & Goodman, C. Value and need as organizing factors
in perception. /. abnorm. soc. Psychol, 1947, 42, 33-44.

38. Bruner, J., & Postman, L. Emotional selectivity in perception and
reaction. /. Pers., 1947, 16, 69-77.

The Structure of Psychoanalytic Theory 169

39. Bruner, J.5 & Postman, L. Tension and tension release as organizing
factors in perception. /. Pers., 1947, 15, 300-308.

40. Bniner, J., & Postman, L. Symbolic value as an organizing factor In
perception. /. soc. PsychoL, 1948, 27, 203-208.

41. Bruner, J., Postman, L., & McGinnies, E. Personal values as selective
factors in perception. /. abnorm. soc. PsychoL, 1948, 43, 142-154.

42. Buehler, K. Die geistige Entwicklung des Kindes. (6th ed.) Jena:
Fischer, 1930.

43. Chein, I. The genetic factor in ahistorical psychology. /. gen.
PsychoL, 1947, 36, 151-172.

44. Delbrueck, M. A physicist looks at biology. Trans. Conn. Acad. Arts
ScL, 1949, 38, 175-190.

45. Deutsch, F. The choice of organ in organ neuroses. Int. J. Psycho-
Anal, 1939, 20, 252-262.

46. Dingle, H. The scientific adventure. London: Pitman, 1952. New
York: Philosophical Library, 1953.

47. Diven, K. Certain determinants In the conditioning of anxiety re-
actions. /. PsychoL, 1937, 3, 219-308.

48. Bollard, J., & Miller, N. E. Personality and psychotherapy: an
analysis in terms of learning, thinking and culture. New York: McGraw-
Hill, 1950.

49. Dunbar, H. F. Emotions and bodily changes. (3d ed.) New York:
Columbia Univer. Press, 1947.

50. Dunbar, H. F. Mind and body: psychosomatic medicine. New York:
Random House, 1947.

51. Eissler, K. R. The effect of the structure of the ego on psychoanalytic
technique. /. Amer. psychoanal. Ass., 1953, 1, 104-143.

52. Ellis, W. D. A sourcebook of Gestalt psychology. New York:
Humanities Press, 1950.

53. Erickson, M. H. Experimental demonstration of the psychopathology
of everyday life. Psychoanal. Quart., 1939, 8, 338-353.

54. Erickson, M. H., & Kubie, L. S. The translation of the cryptic
automatic writing of one hypnotic subject by another in a trance-like dis-
sociated state. Psychoanal Quart., 1940, 9, 51-63.

55. Erickson, M. H., & Kubie, L. S. The successful treatment of a case
of acute hysterical depression by a return under hypnosis to a critical phase
of childhood. Psychoanal Quart., 1941, 10, 583-609.

56. Erikson, E. H. Configurations in play — clinical notes. Psychoanal.
Quart., 1937, 6, 139-214.

57. Erikson, E. H. Observations on Sioux education. /. PsychoL, 1939,
7, 101-156.

58. Erikson, E. H. Problems of infancy and early childhood. In
Cyclopedia of medicine. Philadelphia: Davis, 1940. Pp. 714-730. Also in
G. Murphy & A. Bachrach (Eds.), Outline of abnormal psychology. New
Fork: Modern Library, 1954. Pp. 3-36.

59. Erikson, E. H. Childhood and tradition in two American Indian
tribes. In The psychoanalytic study of the child. Vol. 1. New York: Inter-

170 DAVID RAPAPORT

national Universities Press, 1945. Pp. 319-350. Also (revised) in C. Kluck-
hohn & H. Murray (Eds/), Personality in nature, society and culture. New
York: Knopf, 1948. Pp. 176-203.

60. Erikson, E. H. Ego development and historical change. In The psy-
choanalytic study of the child. Vol. 2. New York: International Uni-
versities Press, 1946. Pp. 359-396.

61. Erikson, E. H. Childhood and society. New York: Norton, 1950.

62. Erikson, E. H. Growth and crises of the "healthy personality." In C.
Kluckhohn & H. Murray (Eds.) , Personality in nature, society and culture.
(2d ed.) New York: Knopf, 1953. Pp. 185-225.

63. Erikson, E. H. The dream specimen of psychoanalysis. /. Amer.
psychoanal Ass., 1954, 2, 5-56. Also in R. P. Knight £ G. R. Friedman
(Eds.), Psychoanalytic psychiatry and psychology, clinical and theoretical
papers. Austen Riggs Center. Vol. 1. New York: International Universities
Press, 1954. Pp. 131-170.

64. Erikson, E. H. Freud's "The origins of psychoanalysis." Int. J.
Psycho-Anal, 1955, 36, 1-15.

65. Erikson, E. H. The first psychoanalyst: crisis and discovery. Yale
Rev., 1956, 46, 40-62.

66. Erikson, E. H. The problem of ego identity. /. Amer. psychoanal.
Ass., 1956, 4, 56-121.

67. Escalona, S. Problems in psycho-analytic research. Int. J. Psycho-
Anal, 1952,32, 11-21.

68. Farber, L., & Fisher, C. An experimental approach to dream psy-
chology through the use of hypnosis. Psychoanal Quart., 1943, 12, 202-216.

69. Fechner, G. T. Einige Ideen zur Schopfungs- und Entwicklungsge-
schichte der Organismen. Leipzig: Breitkopf und Hartel, 1873.

70. Federn, P. Ego psychology and the psychoses. New York: Basic
Books, 1952.

71. Feigl, H. Principles and problems of theory construction in psy-
chology. In Current trends in psychological theory. Pittsburgh: Univer.
Pittsburgh Press, 1951. Pp. 179-213.

72. Fenichel, O. Problems of psychoanalytic technique. New York:
Psychoanalytic Quarterly, Inc., 1941.

73. Fenichel, O. The psychoanalytic theory of neurosis. New York:
Norton, 1945.

74. Fenichel, O. A critique of the death instinct. In [77, first series,
pp. 363-372].

75. Fenichel, O. The ego and the affects. In [77, second series, pp.
215-227].

76. Fenichel, O. Psychoanalytic remarks on Fromm's book, "Escape
from freedom." In [77, second series, pp. 260-277].

77. Fenichel, O. Collected papers. First and second series. New York:
Norton, 1953-54.

78. Ferenczi, S. Stages in the development of the sense of reality. In
Sex in psychoanalysis. New York: Brunner, 1950. Pp. 213-239.

The Structure of Psychoanalytic Theory 171

79. Ferenczi, S. The problem of acceptance of unpleasant ideas. In
Further contributions to the theory and technique of psycho-analysis. Lon-
don: Hogarth, 1950. Pp. 366-379.

80. Ferenczi, S. The psyche as an inhibiting organ. In Further contribu-
tions to the theory and technique of psycho-analysis. London: Hogarth,
1950. Pp. 379-383.

81. Ferenczl, S. Thinking and muscle innervation. In Further contribu-
tions to the theory and technique of psycho-analysis. London: Hogarth,
1950. Pp. 230-232.

82. Finch, G.. & Culler, E. Relation of forgetting to experimental ex-
tinction. Amer. J. PsychoL, 1935, 47, 656-662.

83. Fisher, G. Studies on the nature of suggestion. Parts I and II. /.
Amer. psychoanal. Ass., 1953, 1, 222-255, 406-437.

84. Fisher, G. Dreams and perception: the role of preconscious and
primary modes of perception in dream formation. /. Amer. psychoanal. Ass.,
1954, 2, 389-445.

85. French, T. M. Interrelations between psychoanalysis and the ex-
perimental work of Pavlov. Amer. J. Psychiat., 1933, 12 (89), 1165-1203.

86. French, T. M. Psychogenic factors in asthma. Amer. J. Psychiat..,
1939, 96, 87-101.

87. French, T. M. Goal, mechanism and the integrative field. Psychosom.
Med., 1941, 3, 226-252.

88. French, T. M. The integrative process. Dialectic a, 1951, 5, 246-256.

89. French, T. M. The integration of behavior. Vols. 1 & 2. Chicago:
Univer. Chicago Press, 1952, 1954.

90. French, T. M., & Alexander, F. Psychogenic factors in bronchial
asthma. Parts I and II. Psychosom. Med. Monogr., 1941, 1 (4); 2 (1, 2).

91. Frenkel-Brunswik, E. Motivation and behavior. Genet. Psychol.
Monogr., 1942, 26, 121-265.

92. Frenkel-Brunswik, E. Psychoanalysis and the unity of science. Proc.
Amer. Acad. Arts. ScL, 1954, 80, 271-350.

93. Freud, A. (1936) The ego and the mechanisms of defence. New
York: International Universities Press, 1946.

94. Freud, S. (1887-1902) The origins of psycho-analysis: letters to
Wilhelm Fliess, drafts and notes: 1887-1902. New York: Basic Books, 1954.

95. Freud, S. (1893-1914) Collected papers of . . . . Vol. 1. London:
Hogarth, 1948.

96. Freud, S. (1894) The defence neuro-psychoses. In [95, pp. 59-75].

97. Freud, S. (1896) Further remarks on the defence neuro-psychoses.
In [95, pp. 155-182].

98. Freud, S. (1900) The interpretation of dreams. In The basic writ-
ings of .... New York: Modern Library, 1938. Pp. 179-549.

99. Freud, S. (1904) The psychopathology of everyday life. In The
basic writings of . . . . New York: Modern Library, 1938. Pp. 33-178.

100. Freud, S. (1905) Wit and its relation to the unconscious. In The
basic writings of . . . . New York: Modern Library, 1938. Pp. 631-803.

172 DAVID RAPAPORT

101. Freud, S. (1905) Three essays on the theory of sexuality. London:
Imago, 1949.

102. Freud, S. (1907) Delusion and dream. London: Allen & Unwin,
1921.

103. Freud, S. (1908) The relation of the poet to day-dreaming. In
Collected papers of . . . . Vol. 4. London: Hogarth, 1948. Pp. 173-183.

104. Freud, S. (1910) The antithetical sense of primal words. In Col-
lected papers of . . . . Vol. 4. London: Hogarth, 1948. Pp. 184-191.

105. Freud, S. (1910) Leonardo da Vinci. New York: Random House,
1947.

106. Freud, S. (1910-1919) Papers on technique. In Collected papers
of . . . . Vol. 2. London: Hogarth, 1948. Pp. 285-402.

107. Freud, S. (1911) Psycho-analytic notes upon an autobiographical
account of a case of paranoia (dementia paranoides). In Collected papers
of . . . . Vol. 3. London: Hogarth, 1948. Pp. 387-470.

108. Freud, S. (1911) Formulations regarding the two principles in
mental functioning. In Collected papers of . . . . Vol. 4. London: Hogarth,
1948. Pp. 13-21.

109. Freud, S. (1912) Totem, and taboo. In The basic writings of
.... New York: Modern Library, 1938. Pp. 805-930.

110. Freud, S. (1912) A note on the unconscious in psycho-analysis. In
Collected papers of .... Vol. 4. London: Hogarth, 1948. Pp. 22-29.

111. Freud, S. (1913) The occurrence in dreams of material from fairy-
tales. In Collected papers of . . . . Vol. 4. London: Hogarth, 1948. Pp.
236-243.

112. Freud, S. (1913) The theme of the three caskets. In Collected
papers of . . . . Vol. 4. London: Hogarth, 1948. Pp. 244-256.

113. Freud, S. (1914) The Moses of Michelangelo. In Collected papers
of . . . . Vol. 4. London: Hogarth, 1948. Pp. 257-287.

114. Freud, S. (1914) On narcissism: an introduction. In Collected
papers of . . . . Vol. 4. London: Hogarth, 1948. Pp. 30-59.

115. Freud, S. (1915) Instincts and their vicissitudes. In Collected
papers of . . . . Vol. 4. London: Hogarth, 1948. Pp. 60-83.

116. Freud, S. (1915) Repression. In Collected papers of . . . . Vol. 4.
London: Hogarth, 1948. Pp. 84-97.

117. Freud, S. (1915) The unconscious. In Collected papers of . . . .
Vol. 4. London: Hogarth, 1948. Pp. 98-136.

118. Freud, S. (1916) A mythological parallel to a visual obsession. In
Collected papers of . . . . Vol. 4. London: Hogarth, 1948. Pp. 345-346.

119. Freud, S. (1916) Metapsychological supplement to the theory of
dreams. In Collected papers of . . . . Vol. 4. London: Hogarth, 1948. Pp.
137-151.

120. Freud, S. (1917) Mourning and melancholia. In Collected papers
of . . . . Vol. 4. London: Hogarth, 1948. Pp. 152-170.

121. Freud, S. (1917) A general introduction to psycho-analysis. New
York: Perma Giants, 1949.

The Structure of Psychoanalytic Theory 173

122. Freud, S. (1919). A cMId is being beaten. In Collected papers
of . . . . Vol. 2. London: Hogarth, 1948. Pp. 172-201.

123. Freud, S. (1920) Beyond the pleasure principle. London: Hogarth.
1948.

124. Freud, S. (1921) Group psychology and the analysis of the ego.
London: Hogarth, 1948.

125. Freud, S. (1922) Medusa's head. In Collected papers of . . . .
Vol. 5. London: Hogarth, 1950. Pp. 105-106.

126. Freud, S. (1923) The ego and the id. London: Hogarth, 1947.

127. Freud, S. (1924) Neurosis and psychosis. In Collected papers of
.... Vol. 2. London: Hogarth, 1948. Pp. 250-254.

128. Freud, S, (1924) The loss of reality in neurosis and psychosis. In
Collected papers of . . . . Vol. 2. London: Hogarth, 1948. Pp. 277-282.

129. Freud, S. (1925) Negation. In Collected papers of . . . . Vol. 5.
London: Hogarth, 1950. Pp. 181-185. Also in [268, pp. 338-348].

130. Freud, S. (1925) A note upon the "mystic writing-pad.55 In Col-
lected papers of . . . . Vol. 5. London: Hogarth, 1950. Pp. 175-180. Also
in [268, pp. 329-337].

131. Freud, S. (1926) The problem of anxiety. New York: Psycho-
analytic Quarterly and Norton, 1936.

132. Freud, S. (1927) The future of an illusion. New York: Liveright,
1949.

133. Freud, S. (1928) Dostoevsky and parricide. In Collected papers of
.... Vol. 5. London: Hogarth, 1950. Pp. 222-242.

134. Freud, S. (1928) Humour. In Collected papers of . . . . Vol. 5.
London: Hogarth, 1950. Pp. 215-221.

135. Freud, S. (1928) A religious experience. In Collected papers of
.... Vol. 5. London: Hogarth, 1950. Pp. 243-246.

136. Freud, S. (1930) Civilization and its discontents. London: Hogarth,
1946.

137. Freud, S. (1932) New introductory lectures on psycho-analysis.
New York: Norton, 1933.

138. Freud, S. (1932) Why war? In John Rickman (Ed.), Civilization,
war, and death. London: Hogarth, 1939. Pp. 82-97.

139. Freud, S. (1937) Analysis terminable and interminable. In Col-
lected papers of . . . . Vol. 5. London: Hogarth, 1950. Pp. 316-357.

140. Freud, S. (1938) An outline of psychoanalysis. New York: Norton,
1949.

141. Freud, S. (1939) Moses and monotheism. London: Hogarth, 1939.

142. Fromm-Reichmann, F. Transference problems in schizophrenics.
Psychoanal. Quart., 1939, 8, 412-426.

143. Fromm-Reichmann, F. Notes on the development of treatment of
schizophrenics by psychoanalytic psychotherapy. Psychiatry, 1948, 11, 263-
273.

144. Gardner, R. W. Cognitive styles in categorizing behavior. /. Pers.y
1953, 22, 214^233.

174 DAVID RAPAPORT

145. GUI, M. M. Psychoanalysis and exploratory psychotherapy. /. Amer.
psychoanal Ass., 1954, 2? 771-797.

146. Gill, M. M. The present state of psychoanalytic theory. /. abnorm.
soc. PsychoL, 1959, 58, 1-8.

147. Gill, M. M., & Brenman, M. Problems in clinical research round
table, 1946. Amer. ]. OrthopsychiaL, 1947, 17, 196-230.

148. Gill, M. M., & Brenman, M. Hypnosis and related states: psycho-
analytic studies in regression. New York: International Universities Press,
in press.

149. Gill, M. M., & Rapaport, D. The points of view and assumptions
of metapsychology. Int. J. Psycho-Anal., in press.

150. Glover, E. Basic mental concepts. London: Imago, 1947.

151. Glover, E. The technique of psycho-analysis. (Rev. Ed.) New
York: International Universities Press, 1955.

152. Goldmann, A. E. Studies in vicariousness : degree of motor activity
and the autokinetic phenomenon. Amer. ]. PsychoL, 1953, 66, 613-
617.

153. Gross, A. The Secret Bull Menninger Clin.3 1951, 15, 37-44.

154. Gross, A. Freud3s creative period. Unpublished manuscript.

155. Halverson, H. M. Infant sucking and tensional behavior. /. genet.
PsychoL, 1938, 53, 365-430.

156. Halverson, H. M. Genital and sphincter behavior of the male
infant /. genet. PsychoL, 1940, 56, 95-136.

157. Hartmann, H. Ego psychology and the problem of adaptation,
New York: International Universities Press, 1958. Also, abridged, in [268,
pp. 362-396].

158. Hartmann, H. Psychoanalysis and the concept of health. Int. J.
Psycho- Anal., 1939, 20, 308-321.

159. Hartmann, H. On rational and irrational action. In G. Roheim
(Ed.), Psychoanalysis and the social sciences. Vol. 1. New York: Inter-
national Universities Press, 1947. Pp. 359-392.

160. Hartmann, H. Comments on the psychoanalytic theory of in-
stinctual drives. Psychoanal. Quart., 1948, 17, 368-388.

161. Hartmann, H. Comments on the psychoanalytic theory of the ego.
In The psychoanalytic study of the child. Vol. 5. New York: International
Universities Press, 1950. Pp. 74-96.

162. Hartmann, H. The mutual influences in the development of the
ego and id. In The psychoanalytic study of the child. Vol. 7. New York:
International Universities Press, 1952. Pp. 9-30.

163. Hartmann, H. Contribution to the metapsychology of schizophrenia.
In The psychoanalytic study of the child. Vol. 8. New York: International
Universities Press, 1953. Pp. 177-198.

164. Hartmann, H. Notes on the theory of sublimation. In The psy-
choanalytic study of the child. Vol. 10. New York: International Uni-
versities Press, 1955. Pp. 9-29.

165. Hartmann, H. The development of the ego concept in Freud's
work. Int. J. Psycho- Anal, 1956, 37, 425-438.

The Structure of Psychoanalytic Theory 175

166. Hartmann, H.3 & Kris, E. The genetic approach in psychoanalysis.
In The psychoanalytic study of the child. Vol. L New York: International
Universities Press, 1945. Pp.' 11-29.

167. Hartmann, H., Kris, E.. & Loewenstein, R. M. Comments on the
formation of psychic structure. In The psychoanalytic study of the child.
Vol. 2. New York: International Universities Press, 1946. Pp. 11-38.

168. Hartmann. H., Kris, E., & Loewenstein, R. M. Notes on the theory
of aggression. In The psychoanalytic study of the child. Vols. 3 & 4. New
York: International Universities Press, 1949. Pp. 9-36.

169. Hebb, D. O. The organization of behavior: a neuro psychological
theory. New York: Wiley, 1949.

170. Heron, W. The pathology of boredom. Sci. Amer., 1957, 196, 52-56.

171. Heron, W., Bexton, W. H., & Hebb, D. O. Cognitive effects of a
decreased variation in the sensory environment. Amer. Psychol., 1953, 8,
366.

172. Heron, W., Doane, B. K., & Scott, T. H. Visual disturbances after
prolonged perceptual isolation. Canad. /. PsychoL, 1956, 10, 13-18.

173. Hilgard, E. R. Experimental approaches to psychoanalysis. In
E. Pumpian-Mindlin (Ed.), Psychoanalysis as science. Stanford, Calif.:
Stanford Univer. Press, 1952. Pp. 3-45.

174. Holt, E. B. The Freudian wish and its place in ethics. New York:
Holt, 1915.

175. Holt, E. B. Animal drive and the learning process. New York:
Holt, 1931.

1 76. Holt, R. R. Implications of some contemporary personality theories
for Rorschach rationale. In Klopfer, B., Ainsworth, M., Klopfer, W., & Holt,
R. R., Developments in the Rorschach technique. Vol. 1. Yonkers, N.Y.:
World Book Co., 1954. Pp. 501-560.

177. Holt, R. R. Freud and the scientific method of psychoanalysis. Un-
published manuscript.

178. Holt, R. R. Gauging primary and secondary processes in Rorschach
responses. /. pro]. Tech., 1956, 20, 14-25.

179. Holzman, P. S., & Klein, G. S. Cognitive system-principles of level-
ing and sharpening: individual differences in assimilation effects in visual
time-error. /. PsychoL, 1954, 37, 105-122.

180. Holzman, P. S. The relation of assimilation tendencies in visual,
auditory, and kinesthetic time-error to cognitive attitudes of leveling and
sharpening. /. Pers.3 1954, 22, 375-394.

181. Homey, K. The neurotic personality of our time. New York:
Norton, 1937.

182. Homey, K. New ways in psychoanalysis. New York: Norton, 1939.

183. Hull, C. L. Psychology seminar memoranda, 1934-1937. Unpub-
lished manuscript, on deposit at Yale Univer. Library.

184. Humphrey, G. The conditioned reflex and the Freudian wish. /.
abnorm. soc. PsychoL, 1920, 14, 388-392.

185. Humphrey, G. Education and Freudianism. /. abnorm. soc.
PsychoL, 1921, 15, 350-402.

176 DAVID RAPAPORT

186. Hunt, J. McV. The effects of infant feeding-frustration upon adult
hoarding in the albino rat. /. abnorm. soc, PsychoL, 1941, 36, 338-360.

187. Hunt, J. McV., Schlosberg, H., Solomon, R. L., & Stellar, E.
Studies of the effect of infantile experience on adult behavior in rats. I.
Effects of infantile feeding frustration on adult hoarding. /. comp. physiol.
PsychoL, 1947, 40, 291-304.

188. Huston, P. E., Shakow, D., & Erickson, M. H. A study of hypnot-
ically induced complexes by means of the Luria technique. /. gen. PsychoL,
1934, 11, 65-97.

189. Jacobson, E. Observations on the psychological effect of imprison-
ment on female political prisoners. In K. Eissler (Ed.), Searchlights on
delinquency. New York: International Universities Press, 1949. Pp. 341-368.

190. Jacobson, E. Contribution to the metapsychology of cyclothymic
depression. In P. Greenacre (Ed.), Affective disorders. New York: Inter-
national Universities Press, 1953. Pp. 49-83.

191. Jacobson, E. The affects and their pleasure-unpleasure qualities,
in relation to psychic discharge processes. In R. Loewenstein (Ed.), Drives,
affects, behavior. New York: International Universities Press, 1953. Pp.,
38-66.

192. Jacobson, E. The self and the object world. In The psychoanalytic
study of the child. Vol. 9. New York: International Universities Press, 1954.
Pp. 75-127.

193. Jones, E. The life and work of Sigmund Freud, Vols. 1 & 2. New
York: Basic Books, 1953, 1955.

194. Kardiner, A. The individual and his society: the psycho dynamics
of primitive social organization. New York: Columbia Univer. Press, 1939.

195. Kardiner, A. The psychological frontiers of society. New York:
Columbia Univer. Press, 1945.

196. Kempf, E. J. The autonomic functions and the personality. New
York: Nervous and Mental Disease Publications, 1921.

197. Klein, G. S. Need and regulation. In M. Jones (Ed.), Nebraska
symposium on motivation. Lincoln, Neb.: Univer. Nebraska Press, 1954.
Pp. 224-274.

198. Klein, G. S. Perception, motive, and personality: a clinical perspec-
tive. To be published by Basic Books. See also in J. McCary (Ed.), Psy-
chology of personality: six modern approaches. New York: Logos Press,
1956. Pp. 123-199.

199. Klein, G. S., & Schlesinger, H. J. Perceptual attitudes toward in-
stability. I. Prediction of apparent movement experiences from Rorschach
responses. /. Pers., 1951, 19, 289-302.

200. Klein, G. S., Schlesinger, H. J., & Meister, D. E. The effects of
personal values on perception: an experimental critique. PsychoL Rev,,
1951, 58, 96-112.

201. Klein, G. S., Spence, D. P., Holt, R. R., & Gourevitch, S. Pre-
conscious influences upon conscious cognitive behavior. Amer. PsychoL,
19553 10, 380. (Abstract) Published as Cognition without awareness: I.

The Structure of Psychoanalytic Theory 177

Subliminal Influences upon conscious thought. /. abnorm. soc. PsychoL,
1958. 57, 255-266.

202. Knight, R. P. Borderline States. Bull Menninger Clin.} 1953, 17,
1-12. Also In R. P. Knight & G. R. Friedman fEds.K Psychoanalytic psy-
chiatry and psychology, clinical and theoretical papers. Austen Riggs Center.
Vol. 1. New York: International Universities Press, 1954. Pp. 97-109.

203. Knight, R. P. Management and psychotherapy of the borderline
schizophrenic patient. Bull Menninger Clin., 1953. 17, 139-150. Also in
R. P. Knight & C. R. Friedman (Eds.), Psychoanalytic psychiatry and psy-
chology, clinical and theoretical papers. Austen Riggs Center. Vol. 1. New
York: International Universities Press, 1954. Pp. 110-122.

204. Koffka, K. Principles of Gestalt psychology. New York: Harcourt,
Brace, 1935.

205. Kris, E. Notes on the development and on some current problems
of psychoanalytic child psychology. In The psychoanalytic study of the child.
Vol. 5. New York: International Universities Press, 1950. Pp. 24-46.

206. Kris, E. On preconscious mental processes. Psychoanal Quart.,
1950, 19, 540-560. Also in [268, pp. 474-493: and 207, pp. 303-318].

207. Kris, E. Psychoanalytic explorations in art. New York: International
Universities Press, 1952.

208. Knis, D., Werner, H., & Wapner. S. Studies in vlcariousness : motor
activity and perceived movement. Amer. ]. Psychol, 1953,, 66,, 603-608.

209. Kubie, L. S. Problems in clinical research round table, 1946. Amer.
J. Orthopsychiat., 1947, 17, 196-230.

210. Kubie, L. S. Problems and techniques of psychoanalytic validation
and progress. In E. Pumpian-Mindlin (Ed.), Psychoanalysis as science.
Stanford, Calif.: Stanford Univer. Press, 1952. Pp. 46-124.

211. Leeper, R. W. A motivational theory of emotion to replace
"emotion as disorganized response." Psychol. Rev., 1948, 55, 5-21.

212. Levy, D. M. Experiments on the sucking reflex and social behavior
of dogs. Amer. J. Orthopsychiat., 1934, 4, 203-224.

213. Levy, D. M. On instinct-satiation: an experiment on pecking be-
havior of chickens. /. gen. Psychol, 1938, 18, 327-348.

214. Lewin, B. The psychoanalysis of elation. New York: Norton, 1950.

215. Lewin, K. Gesetz und Experiment in der Psychologic. Symposion,
1927, 1, 375-421.

216. Lewin, K. A dynamic theory of personality. New York: McGraw-
Hill, 1935.

217. Lewin, K. Principles of topological psychology. New York: Mc-
Graw-Hill, 1936.

218. Lewin, K. Psychoanalysis and topological psychology. Bull Men-
ninger Clin., 1937, 1, 202-212.

219. Lewin, K. The conceptual representation and measurement of
psychological forces. Contr. psychol Theor., 1, (4). Durham, N.C.: Duke
Univer. Press, 1938.

220. Lewin, K. Comments concerning psychological forces and energies,
and the structure of the psyche. In [268, pp. 76-94].

178 DAVID RAPAPORT

221. Lewin, K. Intention, will, and need. In [268, pp. 95-153].

222. Lilly, J. C. Discussion in Illustrative strategies for research on
psycho pathology in mental health. Group for the Advancement of Psy-
chiatry, Symposium No. 2, June, 1956. Pp. 13-20.

223. Lilly, J. C. Mental effects of reduction of ordinary levels of physical
stimuli on intact, healthy persons. Psychiat. Res. Reports, 1956, 5, 1-9.

224. Lorand, S. Technique of psychoanalytic therapy. New York: In-
ternational Universities Press, 1946.

225. Lorenz, K. Der Kumpan in der Umwelt des Vogels. Der Art-
genosse als ausloesendes Moment sozialer Verbal tungsweisen. /. Ornith.,
1935, 83, 137-213.

226. McClelland, D. C. The psychology of mental content reconsidered.
Psychol Rev., 1955, 62, 297-302.

227. MacCorquodale, K., & Meehl, P. E. Operational validity of inter-
vening constructs. In M. Marx (Ed.), Psychological theory. New York:
Macmillan, 1951. Pp. 103-111.

228. Maier, N. R. F. Frustration: the study of behavior without a goal.
New York: McGraw-Hill, 1949.

229. Masserman, J. H. Behavior and neurosis: an experimental psycho-
analytic approach to psycho biologic principles. Chicago: Univer. Chicago
Press, 1943.

230. Meng, H. Das Problem der Organpyschose. Int. Z. PsychoanaL,

1934, 20, 439-458.

231. Menkes, A., & Menkes, J. The application of dimensional analysis
to learning theory. Psychol. Rev., 1957, 64, 8-13.

232. Merlan, P. Brentano and Freud. /. Hist. Ideas, 1945, 6, 375-377.

233. Merlan, P. Brentano and Freud: a sequel. /. Hist. Ideas, 1949, 10,
451.

234. Miller, N. E. Studies of fear as an acquirable drive: I. Fear as
motivation and fear-reduction as reinforcement in the learning of new
responses. /. exp. Psychol, 1948, 38, 89-101.

235. Miller, N. E. Theory and experiment relating psychoanalytic dis-
placement to stimulus-response generalization. /. abnorm. soc. Psychol.,
1948, 43, 155-178.

236. Miller, N. E. Comments on theoretical models, illustrated by the
development of a theory of conflict behavior. /. Pers., 1951, 20, 82-100.

237. Miller, N. E., & Bollard, J. Social learning and imitation. New
Haven, Conn.: Yale Univer. Press, 1941.

238. Morgan, C. D., & Murray, H. A. A method for investigating
phantasies: the Thematic Apperception Test. Arch. Neurol. Psychiat.,

1935, 34, 289-306.

239. Mowrer, O. H. Learning theory and personality dynamics. New
York: Ronald, 1950.

240. Munroe, R. L. Schools of psychoanalytic thought. New York:
Dryden, 1955.

241. Murphy, G. Personality: a biosocial approach to origins and struc-
ture. New York: Harper, 1947.

The Structure of Psychoanalytic Theory 179

242. Murray, H. A. The effect of fear upon estimates of the malicious-
ness of other personalities. /. sac. Psycho!., 1933. 4, 310-329.

243. Murray, H. A., et al Explorations in personality. New York: Ox-
ford Univer. Press, 1938.

244. Nachmansohn, M. Concerning experimentally produced dreams.
In [268, pp. 257-287].

245. Nunberg, H. The synthetic function of the ego. In Practice and
theory of psychoanalysis. New York: Nervous and Mental Disease Publica-
tions, 1948. Pp. 120-136.

246. Page, J., & Warkentin, J. Masculinity and paranoia. /. abnorm.
soc. Psychol, 1938, 33, 527-531.

247. Piaget, J. Judgment and reasoning in the child. New York: Har-
court, Brace, 1928.

248. Piaget, J. The child's conception of the world. New York: Har-
court, Brace, 1929.

249. Piaget, J. The child's conception of physical causality. London:
Routledge, 1930.

250. Piaget, J. Children's philosophies. In C. Murchison (Ed.), A hand-
book of child psychology. Worcester, Mass.: Clark Univer. Press, 1931.
Pp. 377-391.

251. Piaget, J. The language and thought of the child. (2d. ed.)
London: Routledge, 1932.

252. Piaget, J. Introduction a Fepistemologie genetique. Vols. 1, 2, & 3.
Paris: Presses Universitaires de France, 1950.

253. Piaget, J. The psychology of intelligence. New York: Harcourt,
Brace, 1950.

254. Piaget, J. The origins of intelligence in children. New York: Inter-
national Universities Press, 1952.

255. Piaget, J. The construction of reality in the child. New York: Basic
Books, 1954.

256. Piaget, J. Playy dreams and imitation in childhood. New York:
Norton, 1951.

257. Poetzl, O. Experimentell erregte Traumbilder in ihren Beziehungen
zum indirekten Sehen. Z. Neurol. Psychiat., 1917, 37, 278-349.

258. Rapaport, D. Emotions and memory. (2nd ed.) New York: Inter-
national Universities Press, 1950.

259. Rapaport, D. Principles underlying projective techniques. Charact.
&Pers.} 1942, 10,213-219.

260. Rapaport, D. Scientific methodology of psychoanalysis. Six lectures
given at the Topeka Institute of Psychoanalysis, 1945. Unpublished manu-
script.

261. Rapaport, D. Principles underlying non-pro jective tests of per-
sonality. Ann. N. Y. Acad. Sd., 1946, 46, 643-652.

262. Rapaport, D. Dynamic psychology and Kantian epistemology.
Paper presented at the staff seminar of the Menninger Foundation School
of Clinical Psychology, 1947. Unpublished manuscript.

180 DAVID RAPAPORT

263. Rapaport, D. Interpersonal relations, communication and psy-
chodynamics. Paper presented at the Menninger Foundation General
Seminar, 1949. Unpublished manuscript.

264. Rapaport, D. Book review: "Cybernetics" by Norbert Wiener.
Psychoanal Quart., 1950, 19, 598-603.

265. Rapaport, D. On the psycho-analytic theory of thinking. Int. J.
Psycho-Anal, 1950, 31, 161-170. Also in R. P. Knight & C. R. Friedman
(Eds.) 5 Psychoanalytic psychiatry and psychology, clinical and theoretical
papers. Austen Riggs Center. Vol. L New York: International Universities
Press, 1954. Pp. 259-273.

266. Rapaport, D. The autonomy of the ego. Bull. Menninger Clin.,
1951, 15, 113-123. Also in R. P. Knight & C. R. Friedman (Eds.), Psycho-
analytic psychiatry and psychology, clinical and theoretical papers. Austen
Riggs Center. Vol. 1. New York: International Universities Press, 1954.
Pp. 248-258.

267. Rapaport, D. The conceptual model of psychoanalysis. /. Pers.,
1951, 20, 56-81. Also in R. P. Knight & C. R. Friedman (Eds.), Psycho-
analytic psychiatry and psychology, clinical and theoretical papers. Austen
Riggs Center. Vol. 1. New York: International Universities Press, 1954.
Pp. 221-247.

268. Rapaport, D. (Ed.) Organization and pathology of thought. New
York: Columbia Univer. Press, 1951.

269. Rapaport, D. Paul Schilder's contribution to the theory of thought-
processes. Int. J. Psycho-Anal, 1951, 32, 291-301. Also, Appendix in
Schilder, P., Medical psychology. New York: International Universities
Press, 1953. Pp. 340-356.

270. Rapaport, D. Consciousness: a psychopathological and psycho-
dynamic view. In Problems of consciousness. Transactions of the Second
Conference, March 19-20. New York: Josiah Macy, Jr. Foundation, 1951.
Pp. 18-57.

271. Rapaport, D. Book review: O. Hobart Mowrer, "Learning theory
and personality dynamics." /. abnorm. soc. PsychoL, 1952, 47, 137-142.

272. Rapaport, D. Book review: J. Bollard and N. E. Miller, "Per-
sonality and psychotherapy: an analysis in terms of learning, thinking, and
culture." Amer. J. Orthopsychiat., 1953, 23, 204-208.

273. Rapaport, D. Discussion in H. Powdermaker (Ed.), Mass com-
munications seminar: proceedings of an interdisciplinary seminar. New
York: Wermer-Gren Foundation, 1953. Pp. 121-128.

274. Rapaport, D. On the psycho-analytic theory of affects. Int. J.
Psycho-Anal, 1953, 34, 177-198. Also in R. P. Knight & C. R. Friedman
(Eds.), Psychoanalytic psychiatry and psychology, clinical and theoretical
papers. Austen Riggs Center. Vol. 1. New York: International Universities
Press, 1954. Pp. 274-310.

275. Rapaport, D. Some metapsychological considerations concerning
activity and passivity. Two lectures given at the staff seminar of the Austen
Riggs Center, 1953. Unpublished manuscript.

The Structure of Psychoanalytic Theory 181

276. Rapaport D. Cognitive structures. In Contemporary approaches to
cognition: a symposium held at the University of Colorado. Cambridge,
Mass.: Harvard Univer. Press, 1957.

277. Rapaport D. The development and the concepts of psychoanalytic
ego-psychology. Twelve seminars given at the Western New England Insti-
tute for Psychoanalysis, 1955. Unpublished manuscript (mimeo.'l .

278. Rapaport, D. An historical survey of psychoanalytic ego psychology.
Bull. Philadelphia Ass. Psychoanal, 1958? 8, 105-120.

279. Rapaport, D. Psychoanalysis and developmental psychology. Lecture
given at Clark Univer., Worcester, Mass., Sept. 21, 1957.

280. Rapaport, D. The theory of ego autonomy: a generalization. Bull.
Menninger Clin., 1958, 22, 13-35.

281. Rapaport, D., Gill, M. M.3 & Schafer, R. Diagnostic psychological
testing. Vols. 1 & 2. Chicago: Yearbook Publishers, 1945, 1946.

282. Red!, F., & Wineman, D. Children who hate. Glencoe, III.: Free
Press, 1951. Reprinted with [283] in The treatment of childhood aggression.
New York: Basic Books, 1957.

283. Redl, F., & Wineman, D. Controls from within. Glencoe, 111.:
Free Press, 1952. Reprinted with [282] in The treatment of childhood
aggression. New York: Basic Books, 1957.

284. Reich, W. Char akter analyse. Vienna: SelbstverJag des Verfassers,
1933.

285. Reider, N. A type of psychotherapy based on psychoanalytic prin-
ciples. Bull. Menninger Clin., 1955, 19, 111-128.

286. Reik, T. From thirty years with Freud. New York: International
Universities Press, 1940.

287. Rich, J. L. Goals and trends of research in geology and geography.
Science, 1948, 107, 581-584.

288. Roffenstein, G. Experiments on symbolization in dreams. In [268,
pp. 249-256].

289. Roheim, G. The origin and function of culture. New York:
Nervous and Mental Disease Monograph, 1943.

290. Roheim, G. (Ed.) Psychoanalysis and the social sciences. Vols. 1-4.
New York: International Universities Press, 1947-1955.

291. Roheim, G. Psychoanalysis and anthropology. New York: Inter-
national Universities Press, 1950.

292. Rorschach, H. Psycho diagnostics: a diagnostic test based on per-
ception. New York: Grune & Stratton, 1942.

293. Rosenzweig, S. The experimental study of psychoanalytic concepts.
Charact. & Pers.3 1937, 6, 61-71.

294. Rosenzweig, S. The experimental study of repression. In H. A.
Murray (Ed.), Explorations in personality. New York: Oxford Univer.
Press, 1938. Pp. 472-490.

295. Sanford, R. N. The effects of abstinence from food upon imaginal
processes: a preliminary experiment /. PsychoL, 1936, 2, 129-136.

296. Sanford, R. N. The effects of abstinence from food upon imaginal
processes: a further experiment. /. PsychoL, 1937, 3, 145-159.

182 DAVID RAPAPORT

297. de Saussure, R. Ueber genetische Psychoiogie und Psychoanalyse.
Imago, 1934, 20, 282-315.

298. Schafer, R. Psychoanalytic interpretation in Rorschach testing:
theory and application. New York: Grune & Stratton, 1954.

299. Schilder, P. Mind: perception and thought in their constructive
aspects. New York: Columbia Univer. Press, 1942.

300. Schilder, P. On the development of thoughts. In [268, pp. 497-
518].

301. Schilder, P. Studies concerning the psychology and symptomatology
of general paresis. In [268, pp. 519-580].

302. Schilder, P. Medical psychology. New York: International Uni-
versities Press, 1953.

303. Schiller, C. H. (Ed. & Trans.) Instinctive behavior: the develop-
ment of a modern concept. New York: International Universities Press,
1957.

304. Schlesinger, H. J. Cognitive attitudes in relation to susceptibility
to interference./. Pers., 1954, 22, 354-374.

305. Schroedinger, E. What is life? New York: Macmillan, 1947.

306. Schroetter, K. Experimental Dreams. In [268, pp. 234-248].

307. Sears, R. R. Functional abnormalities of memory with special
reference to amnesia. Psychol. Bull, 1936, 33, 229-274.

308. Sears, R. R. Survey of objective studies of psychoanalytic concepts.
New York: Soc. Sci. Res. Council, Bulletin No. 51, 1943.

309. Shakow, D., & Rapaport, D. The influence of Freud on psychology.
New York : International Universities Press, in press.

310. Silberer, H. Report on a method of eliciting and observing certain
symbolic hallucination-phenomena. In [268, pp. 195-207].

311. Simpson,, G. G. Quantitative zoology: numerical concepts in the
study of recent and fossil animals. New York: McGraw-Hill, 1939.

312. Simpson, G. G. Tempo and mode in evolution. New York: Colum-
bia Univer. Press, 1949.

313. Sullivan, H. S. Conceptions of modern psychiatry. Washington,
B.C.: William Alanson White Psychiatric Foundation, 1947.

314. Sullivan, H. S. The interpersonal theory of psychiatry. New York:
Norton, 1953.

315. Tinbergen, H. The study of instinct. Oxford: Clarendon Press,
1951.

316. Tolman, E. C. The intervening variable. In M. Marx (Ed.), Psy-
chological theory. New York: Macmillan, 1951. Pp. 87-102.

317. Travell, J., & Bigelow, N. H. Role of somatic trigger areas in
the patterns of hysteria. Psychosom. Med.3 1947, 9, 353-363.

318. Troland, L. T. The fundamentals of human motivation. Princeton,
N.J.: VanNostrand, 1928.

319. Varendonck, J. The psychology of daydreams. New York: Mac-
millan, 1921. Also (abridged) in [268, pp. 451-473].

320. Waelder, R. The principle of multiple function: observations on
over-determination. PsychoanaL Quart.,, 1936, 5, 45-62.

Tht Struct u re of Psychoanalytic Thtury 183

321. Weiss, E.. & English, O. S. P-^cJ-^o-^stic rr^ditiKf: ttic clinical

application of psycho pat fio!:gy ij g^r.crcl mfdiccl problem \ Philadelphia:
Saunders, 1943.

322. Werner. H. Comparative psychology of mental development. (3d
ed.) New York: Internationa! Universities Press.. 1957.

323. Werner, BL & Wapner, S. Toward a general theory of perception.
PsychoL Rev., 1952, 593 324-338.

324. Wexler, M. The structural problem in schizophrenia: therapeutic
implications. Int. /. Psycho-Anal., 1951, 325 157-166.

325. Wexler, M. The structural problem in schizophrenia: the role of
the internal object. Bull. Menninger Clin.3 1951, 15, 221-234.

326. Wiener, N. Cybernetics. New York: Wiley. 1948.

327. Wittels, F. Sigmund Freud: his personality y his teaching and his
school New York: Dodd, Mead, 1924.

328. Wolff, P. H. Piaget's genetic psychology and its relation to psy-
choanalysis. Unpublished manuscript.

329. Woodworth, R. S. Dynamic psychology. New York: Columbia
Univer. Press, 1925.

330. ZetzeL E. The depressive position. In P. Greenacre (Ed.), Affec-
tive disorders. New York: International Universities Press, 1953. Pp. 84—116.

A THEORY OF THERAPY, PERSONALITY,
AND INTERPERSONAL RELATIONSHIPS,
AS DEVELOPED IN THE
CLIENT-CENTERED FRAMEWORK

CARL R. ROGERS

University of Wisconsin

Introduction {1} 185

The soil of the theory 185

Some basic attitudes 188

The General Structure of Our Systematic Thinking {2 +, 3} 192

Definitions of constructs 194

A digression on the case history of a construct {3-f }• 200

I. A Theory of Therapy and Personality Change {2+, 6, 8, 9} . . . . 212

Conditions of the therapeutic process 213

The process of therapy 216

Outcomes in personality and behavior 2'18

Comments on the theory of therapy 220

Specification of functional relationships 220

Some conclusions regarding the nature of the individual 220

II. A Theory of Personality {2 +, 6, 9} 221-

Postulated characteristics of the human infant 222

The development of the self 223

The need for positive regard 223

The development of the need for self-regard 224

The development of conditions of worth 224

The development of incongruence between self and experience . . . 226

The development of discrepancies in behavior 227

The experience of threat and the process of defense 227

The process of breakdown and disorganization 228

The process of reintegration 230

Specification of functional relationships in the theory of personality . . 231

Evidence 232-

III. A Theory of the Fully Functioning Person {2 +5 6} 234

184

Therapy, Personality, and Interpersonal Relationships 185

IV. A Theory of Interpersonal Relationship {2-r, 6} 235

V. Theories of Application {2 -r, 6} 241

Family life 241

Education and learning 241

Group leadership 24^

Group tension and conflict 242

The Theoretical System in a Context of Research 244

The bases of stimulation of research {8} 245

The problem of measurement and quantification {5 > 246

Incompatible evidence {9, 7} 247

A continuing program of theory and research {11,7} 249

Immediate strategy of development {12} 250

Conclusion 252

References 252

INTRODUCTION

Being one who has deprecated the use of compulsion as a means of
altering personality and behavior, it is no doubt singularly appropriate
that I should be forced to acknowledge the value of the gentle com-
pulsion of a formal request. For some time I had recognized the need
of a more adequate and more up-to-date statement of the theories which
have been developing in the group associated with client-centered
therapy. This might well have remained in the realm of good intentions,
had it not been for the formal request from the American Psychological
Association, in connection with its Study of the Status and Development
of Psychology in the United States, to prepare a systematic statement
of this developing theory. To join with others who were endeavoring to
formulate their own theories and to use, so far as possible, a common
outline — this seemed to be both an obligation and an opportunity which
could not be refused. It is this softly voiced but insistent pressure from
my colleagues which has caused me to write the following pages now,
rather than at some later date. For this pressure I am grateful.

The soil of the theory. No theory can be adequately understood
without some knowledge of the cultural and personal soil from which
it springs. Consequently I am pleased that the first item of the suggested
outline requests a thorough discussion of background factors. This means,
I fear, that I must take the reader through some autobiographical ma-
terial since, although the client-centered orientation has become very
much of a group enterprise in every respect, I, as an individual, carry a
considerable responsibility for its initiation and for the beginning
formulation of its theories. I shall, therefore, mention briefly some
cultural influences and personal experiences which may or may not

CARL R. ROGERS

186

have relevance to the theory itself. I shall not attempt to evaluate these
influences, since 1 am probably a poor judge of the part they have
played.

I lived my childhood as a middle child in a large, close-knit family,
where hard work and a highly conservative (almost fundamentalist)
Protestant Christianity were about equally revered. When the family
moved to a farm at the time I was twelve, I became deeply interested
and involved in scientific agriculture. The heavy research volumes I read
on my own initiative in the next few years regarding feeds and feeding,
soils, animal husbandry, and the like, instilled in me a deep and abiding
respect for the scientific method as a means of solving problems and
creating new advances in knowledge. This respect was reinforced by rny
first years in college, where I was fond of the physical and biological
sciences. In my work in history I also realized something of the satis-
factions of scholarly work.

Having rejected the family views of religion, I became interested in
a more modern religious viewpoint and spent two profitable years in
Union Theological Seminary, which at that time was deeply committed
to a freedom of philosophical thought which respected any honest at-
tempt to resolve significant problems, whether this led into or away from
the church. My own thinking lead me in the latter direction, and I
moved "across the street" to Teachers College, Columbia University.
Here I was exposed to the views of John Dewey, not directly, but
through William H. KilpatricL I also had my first introduction to
clinical psychology in the warmly human and common-sense approach
of Leta Hollingworth. There followed a year of internship at the In-
stitute for Child Guidance, then in its chaotic but dynamic first year of
existence. Here I gained much from the highly Freudian orientation of
most of its psychiatric staff, which included David Levy and Lawson
Lowrey. My first attempts at therapy were carried on at the Institute.
Because I was still completing my doctorate at Teachers College, the
sharp incompatibility of the highly speculative Freudian thinking of the
Institute with the highly statistical and Thorndikean views at Teachers
College was keenly felt.

There followed twelve years in what was essentially a community
child guidance clinic in Rochester, New York. This was a period of
comparative isolation from the thinking of others. The psychology de-
partment of the University of Rochester was uninterested in what we
were doing because our work was not, in its opinion, in the field of
psychology. Our colleagues in the social agencies, schools, and courts
knew little and cared less about psychological ideologies. The only
element which carried weight with them was the ability to get results
in working with maladjusted individuals. The staff was eclectic, of diverse

Therapy, Personality, and Interpersonal Relationships 187

background, and our frequent and continuing discussion of treatment
methods was based on our practical even-day working experience with
the children, adolescents, and adults who were our clients. It was the
beginning of an effort, which has had meaning for rue ever since, to
discover the order which exists in our experience of working with people.
The volume on the Clinical Treatment of the Problem Child was one
outcome of this effort.

During the second half of this period there were several individuals
who brought Into our group the controversial therapeutic views of Otto
Rank and the Philadelphia group of social workers and psychiatrists
whom he had Influenced. Personal contact with Rank was limited to
a three-day institute we arranged; nevertheless his thinking had a very
decided impact on our staff and helped me to crystallize some of the
therapeutic methods we were groping toward. For by this time I was
becoming more competent as a therapist, and beginning to sense a dis-
coverable orderliness in this experience, an orderliness which was in-
herent in the experience, and (unlike some of the Freudian theories
which had grown so far from their original soil) did not have to be
imposed on the experience.

Though I had earned on some part-time university teaching through-
out the Rochester years, the shift to a faculty position at Ohio State
University w^as a sharp one. I found that the emerging principles of
therapy, which I had experienced largely on an implicit basis, were by
no means clear to well-trained, critically minded graduate students. I
began to sense that what I was doing and thinking In the clinical field
was perhaps more of a new pathway than I had recognized. The paper
I presented to the Minnesota chapter of Psi Chi in December, 1940,
(later chapter 2 of Counseling and Psychotherapy) was the first con-
scious attempt to develop a relatively new line of thought. Up to that
time I had felt that my writings were essentially attempts to distill out
more clearly the principles which "all clinicians" were using.

The new influence at Ohio State, which continued to be felt in my
years at Chicago, was the impact of young men and women — intellectu-
ally curious, often theoretically oriented, eager to learn from experience
and to contribute through research and theory to the development of a
field of knowledge. Through their mistakes as well as their successes in
therapy, through their research studies, their critical contributions, and
through our shared thinking, have come many of the recent develop-
ments in this orientation.

In the past decade at the University of Chicago the new elements
which stand out most sharply are the opportunity for and the encourage-
ment of research, the inclusion of graduate students from education,
theology, human development, sociology, industrial relations, as well as

188 CARL R. ROGERS

psychology, in the ramified activities of the Counseling Center, and the
creative thinking of my faculty colleagues, especially those connected
with the Center.

The persistent influence which might not be fully recognized, because
it is largely implicit in the preceding paragraphs, is the continuing
clinical experience with individuals who perceive themselves, or are
perceived by others to be, in need of personal help. Since 1928, for a
period now approaching thirty years, I have spent probably an average
of 15 to 20 hr per week, except during vacation periods, in endeavoring
to understand and be of therapeutic help to these individuals. To me,
they seem to be the major stimulus to my psychological thinking. From
these hours, and from my relationships with these people, I have drawn
most of whatever insight I possess into the meaning of therapy, the
dynamics of interpersonal relationships, and the structure and function-
ing of personality.

Some basic attitudes. Out of this cultural and personal soil have
grown certain basic convictions and attitudes which have undoubtedly
influenced the theoretical formulation which will be presented. I will
endeavor to list some of these views which seem to me relevant :

1. I have come to see both research and theory as being aimed
toward the inward ordering of significant experience. Thus research
is not something esoteric, nor an activity in which one engages to gain
professional kudos. It is the persistent, disciplined effort to make sense
and order out of the phenomena of subjective experience. Such effort is
justified because it is satisfying to perceive the world as having order and
because rewarding results often ensue when one understands the orderly
relationships which appear to exist in nature. One of these rewarding
results is that the ordering of one segment of experience in a theory im-
mediately opens up new vistas of inquiry, research, and thought, thus
leading one continually forward.

Thus the primary reason for research and systematic theory in the
field of therapy is that it is personally dissatisfying to permit the cumulat-
ing experiences of therapeutic hours to remain as a conglomeration of
more or less isolated events. It feels as though there is an order in these
events. What could it be? And of any hunch regarding the inherent
order, it is necessary to ask the question, is this really true, or am I
deceiving myself? Thus slowly there is assembled a body of facts, and
systematic constructs to explain those facts, which have as their basic
function the satisfaction of a need for order which exists in me.

(I have, at times, carried on research for purposes other than the
above to satisfy others, to convince opponents and sceptics, to gain
prestige, and for other unsavory reasons. These errors in judgment and
activity have only deepened the above positive conviction. )

Therapy, Personality, and Interpersonal Relationships 189

2. It is my opinion that the type of understanding which we call
science can begin anywhere, at any level of sophistication. To observe
acutely, to think carefully and creatively — these activities, not the
accumulation of laboratory instruments, are the beginnings of science.
To observe that a given crop grows better on the rocky hill than in the
lush bottom land, and to think about this observation, is the start of
science. To notice that most sailors get scurvy but not those who have
stopped at islands to pick up fresh fruit is a similar start. To recognize
that, when a person's views of himself change, his behavior changes
accordingly, and to puzzle over this, is again the beginning of both
theory7 and science. I voice this conviction in protest against the attitude,
which seems too common in American psychology, that science starts in
the laboratory or at the calculating machine.

3. A closely related belief is that there is a natural history of science
— that science, in any given field, goes through a patterned course of
growth and development. For example, it seems to me right and natural
that in any new field of scientific endeavor the observations are gross,
the hypotheses speculative and full of errors, the measurements crude.
More important, I hold the opinion that this is just as truly science as the
use of the most refined hypotheses and measurements in a more fully
developed field of study. The crucial question in either case is not the
degree of refinement but the direction of movement. If in either instance
the movement is toward more exact measurement, toward more clear-
cut and rigorous theory and hypotheses, toward findings which have
greater validity and generality, then this is a healthy and growing science.
If not, then it is a sterile pseudo science, no matter how exact its
methods. Science is a developing mode of inquiry, or it is of no par-
ticular importance.

4. In the invitation to participate in the APA study, I have been
asked to cast our theoretical thinking in the terminology of the in-
dependent-intervening-dependent variable, in so far as this is feasible.
I regret that I find this terminology somehow uncongenial. I cannot
justify my negative reaction very adequately, and perhaps it is an
irrational one, for the logic behind these terms seems unassailable. But
to me the terms seem static — they seem to deny the restless, dynamic,
searching, changing aspects of scientific movement. There is a tendency
to suppose that a variable thus labeled, remains so, which is certainly
not true. The terms also seem to me to smack too much of the laboratory,
where one undertakes an experiment de novo, with everything under
control, rather than of a science which is endeavoring to wrest from
the phenomena of experience the inherent order which they contain.
Such terms seem to be more applicable to the advanced stages of
scientific endeavor than to the beginning stages.

190 CARL R. ROGERS

Please do not misunderstand. I quite realize that after the fact, any
research investigation, or any theory constructed to relate the discovered
facts, should be translatable into the language of independent and
dependent variables or there is something wrong with the research or
theory. But the terms seem to me better adapted to such autopsies than
to the living physiology of scientific work in a new field,

5. It should be quite clear from the foregoing that the model of
science which I find most helpful is not taken from the advanced stages
of theoretical physics. In a field such as psychotherapy or personality
the model wiiich seems more congenial to me would be taken from the
much earlier stages of the physical sciences. I like to think of the dis-
covery of radioactivity by the Curies. They had left some pitchblende
ore, which they were using for some purpose or other, in a room where
they stored photographic plates. They discovered that the plates had
been spoiled. In other words, first there was the observation of a
dynamic event. This event might have been due to a multitude of
causes. It might have been a flaw in the manufacture of the plates. It
might have been the humidity, the temperature, or any one of a dozen
other things. But acute observation and creative thinking fastened on
a hunch regarding the pitchblende, and this became a tentative hypoth-
esis. Crude experiments began to confirm the hypothesis. Only slowly
was it discovered that it was not the pitchblende, but a strange element
in the pitchblende which was related to the observed effect. Meanwhile
a theory had to be constructed to bring this strange phenomenon into
orderly relationship with other knowledge. And although the theory in its
most modest form had to do with the effect of radium on photographic
plates, in its wider and more speculative reaches it was concerned with
the nature of matter and the composition of the universe. By present-day
standards in the physical sciences, this is an example of a primitive stage
of investigation and theory construction. But in the fields in which I am
most deeply interested I can only hope that we are approaching such a
stage. I feel sure that we are not beyond it.

6. Another deep-seated opinion has to do with theory. I believe that
there is only one statement which can accurately apply to all theories —
from the phlogiston theory to the theory of relativity, from the theory
I will present to the one which I hope will replace it in a decade — and
that is that at the time of its formulation every theory contains an un-
known (and perhaps at that point an unknowable) amount of error
and mistaken inference. The degree of error may be very great, as in
the phlogiston theory, or small, as I imagine it may be in the theory
of relativity, but unless we regard the discovery of truth as a closed and
finished book, then there will be new discoveries which will contradict
the best theories which we can now construct.

Therapy y Personality, and Interpersonal Relationships 191

To me this attitude Is very Important, for I am distressed at the
manner In which small-caliber minds Immediately accept a theory —
almost any theory — as a dogma of truth. If theory could be seen for what
it Is — a fallible, changing attempt to construct a network of gossamer
threads which will contain the solid facts — then a theory would serve
as it should, as a stimulus to further creative thinking.

I am sure that the stress I place on this grows In part out of my
regret at the history of Freudian theory. For Freud, It seems quite clear
that his highly creative theories were never more than that. He kept
changing, altering, revising, giving new meaning to old terms — always
with more respect for the facts he observed than for the theories he
had built. But at the hands of insecure disciples (so It seems to me),
the gossamer threads became Iron chains of dogma from which dynamic
psychology Is only recently beginning to free Itself. I feel that every
formulation of a theory contains this same risk and that, at the time
a theory is constructed, some precautions should be taken to prevent it
from becoming dosma.

o o

7. I share with many others the belief that truth Is unitary, even
though we will never be able to know this unity. Hence any theory,
derived from almost any segment of experience. If It wrere complete and
completely accurate, could be extended Indefinitely to provide meaning
for other very remote areas of experience. Tennyson expressed this In
sentimental fashion in his "Flower In the Crannied Wall." I too believe
that a complete theory of the Individual plant would show us "what
God and man is."

The corollary, however, is of equal importance and Is not so often
stated. A slight error in a theory may make little difference in providing
an explanation of the observed facts out of which the theory grew.
But wrhen the theory Is projected to explain more remote phenomena,
the error may be magnified, and the inferences from the theory may be
completely false. A very slight error in the understanding of Tennyson's
flower may give a grossly false understanding of man. Thus every theory
deserves the greatest respect In the area from which it was drawn from
the facts and a decreasing degree of respect as it makes predictions in
areas more and more remote from its origin. This is true of the theories
developed by our owrn group.

8. There is one other attitude which I hold, which I believe has
relevance for the proper evaluation of any theory I might present. It is
my belief in the fundamental predominance of the subjective. Man lives
essentially in his own personal and subjective world, and even his most
objective functioning, in science, mathematics, and the like, is the result
of subjective purpose and subjective choice. In relation to research and
theory, for example, It is my subjective perception that the machinery of

192 CARL R. ROGERS

science as we know it — operational definitions, experimental method,
mathematical proof — is the best way of avoiding self-deception. But I
cannot escape the fact that this is the way it appears to me, and that had
I lived two centuries ago, or if I were to live two centuries in the
future, some other pathway to truth might seem equally or more valid.
To put it more briefly, it appears to me that though there may be such
a thing as objective truth, I can never know it; all I can know is that
some statements appear to me subjectively to have the qualifications of
objective truth. Thus there is no such thing as Scientific Knowledge;
there are only individual perceptions of what appears to each person
to be such knowledge.

Since this is a large and philosophical issue, not too closely related to
what follows, I shall not endeavor to state it more fully here but refer
any who are interested to an article in which I have tried to expound
this view somewhat more fully [67]. I mention it here only because
it is a part of the context in which my theoretical thinking has developed.

THE GENERAL STRUCTURE OF OUR SYSTEMATIC THINKING

Before proceeding to the detailed statement of some of our theoretical
\iews, I believe it may be helpful to describe some of the interrelation-
ships between various portions of our theoretical formulations.

The earliest portion, most closely related to observed fact, most
heavily supported by evidence, is the theory of psychotherapy and
personality change which was constructed to give order to the phenomena
of therapy as we experienced it.

In this theory there were certain hypotheses regarding the nature
of personality and the dynamics of behavior. Some of these were ex-
plicit, some implicit. These have been developed more fully into a
theory of personality. The purpose has been to provide ourselves with a
tentative understanding of the human organism and its developing
dynamics — an attempt to make sense of this person who comes to us in
therapy.

Implicit in the theories of therapy and of personality are certain
hypotheses regarding the outcomes of therapy — hence, hypotheses re-
garding a more socially constructive or creative individual. In the last
few years we have endeavored to spell out the picture of the theoretical
end point of therapy, the maximally creative, self-actualizing, or fully
functioning person.

In another direction, our understanding of the therapeutic rela-
tionship has led us to formulate theoretical statements regarding all
interpersonal relationships, seeing the therapeutic relationship simply

Therapy, Personality, and Interpersonal Relationships 193

as one special case. This is a very new and tentative development, which
we believe has promise.

Finally, it has seemed that if our \iews of therapy have any validity
they have application in all those fields of human experience and en-
deavor which involve (a] interpersonal relationships and fb] the aim
or potentiality of development or change in personality and behavior.

n. A THEORY

of

PERSONALITY
H.A 1,2,3,4,5,6
B 1,2
C 1

D 1,2,3,4
E 1,2,3
F 1,2,3
G 1

H 1,2,3,4
I 1,2,3,4
J 1,2,3,

I. A THEORY OF THERAPY

The nature of the human organism

THEORETICAL IMPLICATIONS FOR VARIOUS HUMAN ACTIVITES

3E 3ZE "VTT vi n

FAMILY EDUCATION GROUP GROUP

LIFE LEARNING LEADERSHIP CONFLICT

FlG. I

Consequently a cluster of partially developed theories exists in relation
to such fields as family life, education, group leadership, and situations
of group tension and conflict.

The accompanying chart may help the reader to see and understand
these relationships between different aspects of our theories. It should
be clear that the chart reads from the center, and that the developments
have taken place in the four directions indicated. It should also be
remembered that the possibility of magnification of error in the theory
increases as one goes out from the center. By and large, there is less

194 CARL R. ROGERS

evidence available in these peripheral areas than in the center. Entered
in the chart are the identifying numbers of the various propositions
which follow, so that in reading any specific portion of the theory the
reader may refer back to see its organic relationship to other parts of
the theoretical structure.

Before proceeding to set forth something of the theories themselves,
I should like gratefully to stress the extent to which this is basically a
group enterprise. I have drawn upon specific written contributions to
theory made by Victor Raimy, Richard Hogan, Stanley Standal, John
Butler, and Thomas Gordon. Many others have contributed to my
thinking in ways known and unknown, but I would particularly like to
mention the valuable influence of Oliver Bown, Desmond Cartwright,
Arthur Combs, Eugene Gendlin, A. H. Maslow, Julius Seeman, John
Shlien, and Donald Snygg on the theories which I am about to present.
Yet these individuals are by no means to be held responsible for what
follows, for their own attempts to order experience have often led them
into somewhat different channels of thinking.

Definitions of constructs. In the development of our theories various
systematic constructs have emerged, gradually acquiring sharper and
more specific meaning. Also terms in common usage have gradually
acquired somewhat specialized meanings in our theoretical statements.
In this section I have endeavored to define, as rigorously as I am able,
these constructs and terms. These definitions supply the means by which
the theory may be more accurately understood.

In this section one will find first a numbered list of all of the con-
structs defined, grouped in related clusters. There are eleven of these
clusters, each with a focal concept. If these focal concepts are under-
stood, the understanding of each of the related terms should not be
difficult, since each of the constructs within a group has a close and
meaningful relationship to the others.

Following the list one will find each of the constructs in the order
numbered. Each is defined, and explanatory comment is often added.

In connection with one cluster of concepts, those having to do with
the self, there is a long digression giving the "case history" of the develop-
ment of that construct. This is intended to illustrate the way in which
most of the constructs in this theoretical system have been developed, not
as armchair constructs but out of a continuing interplay between thera-
peutic experience, abstract conceptualizing, and research using opera-
tionally defined terms.

It is quite possible that such a section, devoted entirely to definitions,
will prove dull reading. The reader may prefer to go at once to the
theory of therapy in the following section, where he will find each defined
term printed in italics. He may then refer back to this section for the
exact meaning of each such term.

Therapy, Personality, and Interpersonal Relationships 195
Grouping of Definitions

Actualizing tendency and related construct

1. Actualizing tendency

2. Tendency toward self-actualization
Experience and related constructs

3. Experience f/noun)

4. Experience (verb)

5. Feeling, Experiencing a feeling
Awareness and related constructs

6. Awareness, Symbolization, Consciousness

7. Availability to awareness

8. Accurate symbolization

9. Perceive, Perception

10. Subceive, Subception
Self and related constructs

11. Self-experience

12. Self, Concept of self, Self-structure

13. Ideal self
Incongruence and related constructs

14. Incongruence between self and experience

15. Vulnerability

16. Anxiety

17. Threat

18. Psychological maladjustment
The response to threat

19. Defense, Defensiveness

20. Distortion in awareness, Denial to awareness

21. Intensionality
Congruence and related constructs

22. Congruence of self and experience

23. Openness to experience

24. Psychological adjustment

25. Extensionality

26. Mature, Maturity

Unconditional positive regard and related constructs

27. Contact

28. Positive regard

29. Need for positive regard

30. Unconditional positive regard
- 31. Regard complex

32. Positive self-regard

33. Need for self-regard

34. Unconditional self-regard

196 CARL R. ROGERS

Conditions of worth

35. Conditions of worth
Constructs related to valuing

36. Locus of evaluation

37. Organismic valuing process
Constructs related to source of knowledge

38. Internal frame of reference

39. Empathy

40. External frame of reference

1. Actualizing tendency. This is the inherent tendency of the organ-
ism to develop all its capacities in ways which serve to maintain or en-
hance the organism. It involves not only the tendency to meet what
Maslow [45] terms "deficiency needs" for air, food, water, and the like,
but also more generalized activities/It involves development toward the
differentiation of organs and of functions, expansion in terms of growth,
expansion of effectiveness through the use of tools, expansion, and en-
hancement through reproduction. It is development toward autonomy
and away from heteronomy, or control by external forces. Angyal's state-
ment [2] could be used as a synonym for this term: "Life is an auton-
omous event which takes place between the organism and the en-
vironment. Life processes do not merely tend to preserve life but tran-
scend the momentary status quo of the organism, expanding itself con-
tinually and imposing its autonomous determination upon an ever in-
creasing realm of events."

It should be noted that this basic actualizing tendency is the only
motive which is postulated in this theoretical system. It should also be
noted that it is the organism as a whole, and only the organism as a
whole, which exhibits this tendency. There are no homunculi, no other
sources of energy or action in the system. The self, for example, is an
important construct in our theory, but the self does not "do" anything.
It is only one expression of the general tendency of the organism to
behave in those ways which maintain and enhance itself.

It might also be mentioned that such concepts of motivation as
are termed need-reduction, tension-reduction, drive-reduction, are in-
cluded in this concept. It also includes, however, the growth motivations
which appear to go beyond these terms: the seeking of pleasurable ten-
sions, the tendency to be creative, the tendency to learn painfully to
walk when crawling would meet the same needs more comfortably.

2. Tendency toward self -actualization. Following the development
of the self-structure, this general tendency toward actualization ex-
presses itself also in the actualization of that portion of the experience
of the organism which is symbolized in the self. If the self and the total

Therapy, Personality, and Interpersonal Relationships 197

experience of the organism are relatively congruent, then the actualizing
tendency remains relatively unified. If self and experience are Incon-
gnient, then the general tendency to actualize the organism may work
at cross purposes with the subsystem of that motive, the tendency to
actualize the self.

This definition will be better understood when various of Its terms —
self, Incongruence, etc. — are defined. It is given here because it Is a
sub-aspect of motivation. It should perhaps be reread after the other
terms are more accurately understood.

3. Experience (noun). This term is used to include all that Is going
on within the envelope of the organism at any given moment wrhlch is
potentially available to awareness. It includes events of which the in-
dividual is unaware, as well as all the phenomena which are in con-
sciousness. Thus it includes the psychological aspects of hunger, even
though the individual may be so fascinated by his work or play that he
is completely unaware of the hunger; it includes the impact of sights
and sounds and smells on the organism, even though these are not in
the focus of attention. It includes the influence of memory and past
experience, as these are active in the moment, in restricting or broaden-
ing the meaning given to various stimuli. It also includes all that is
present in immediate awareness or consciousness. It does not Include
such events as neuron discharges or changes in blood sugar, because
these are not directly available to awareness. It is thus a psychological,
not a physiological definition.

Synonyms are "experiential field," or the term "phenomenal field" as
used by Snygg and Combs, which also covers more than the phenomena
of consciousness. I have in the past used such phrases as "sensory and
visceral experiences" and "organic experiences" in the attempt to convey
something of the total quality of this concept.

It is to be noted that experience refers to the given moment, not to
some accumulation of past experience. It is believed that this makes the
operational definition of experience, or of an experience, which is a
given segment of the field, more possible.

4. Experience (verb). To experience means simply to receive in
the organism the impact of the sensory or physiological events which are
happening at the moment.

Often this process term is used in the phrase "to experience in aware-
ness" which means to symbolize in some accurate form at the conscious
level the above sensory or visceral events. Since there are varying de-
grees of completeness in symbolization, the phrase is often "to experience
more fully in awareness," thus indicating that it is the extension of this

198 CARL R. ROGERS

process toward more complete and accurate symbolization to which
reference is being made.

5. Peeling, Experiencing a feeling. This is a term which has been
heavily used in writings on client-centered therapy and theory. It denotes
an emotionally tinged experience, together with its personal meaning.
Thus it includes the emotion but also the cognitive content of the mean-
ing of that emotion in its experiential context. It thus refers to the
unity of emotion and cognition as they are experienced inseparably
in the moment. It is perhaps best thought of as a brief theme of ex-
perience, carrying with it the emotional coloring and the perceived
meaning to the individual. Examples would include "I feel angry at
myself," "I feel ashamed of my desires when I am with her," "For the
first time, right now, I feel that you like me." This last is an example of
another phenomenon which is relevant to our theory, and which has
been called experiencing a feeling fully, in the immediate present. The
individual is then congruent in his experience (of the feeling), his aware-
ness (of it) 3 and his expression (of it) .

6. Awareness, Symbolization, Consciousness. These three terms are
defined as synonymous. To use AngyaPs expression, consciousness (or
awareness) is the symbolization of some of our experience. Aware-
ness [is thus seen as the symbolic representation (not necessarily in verbal
symbols) of some portion of our experience. This representation may
have varying degrees of sharpness or vividness, from a dim awareness
of something existing as ground, to a sharp awareness of something
which is in focus as figure.

7. Availability to awareness. When an experience can be symbolized
freely, without defensive denial and distortion, then it is available to
awareness.

8. Accurate symbolization. The symbols which constitute our aware-
ness do not necessarily match, or correspond to, the "real" experience,
or to "reality." Thus the psychotic is aware of (symbolizes) electrical
impulses in his body which do not seem in actuality to exist. I glance
up quickly and perceive a plane in the distance, but it turns out to be
a gnat close to my eye. It seems important to distinguish between those
awarenesses which, in common-sense terms, are real or accurate and
those which are not. But how can this be conceptualized if we are trying
to think rigorously?

The most adequate way of handling this predicament seems to me
to be to take the position of those who recognize that all perception
(and I would add, all awareness) is transactional in nature, that it is
a construction from our past experience and a hypothesis or prognosis
for the future. Thus the examples given are both hypotheses which

Therapy, Personality,, and Interpersonal Relationships 199

can be checked. If I brush at the gnat and It disappears, it increases the
probability that what I was aware of was a gnat and not a plane. If
the psychotic were able to permit himself to check the electric currents
in Ms body, and to see whether they have the same characteristics as
other electric currents, he would be checking the hypothesis implicit in
his awareness. Hence when we speak of accurate symbolization in aware-
ness, we mean that the hypotheses implicit in the awareness will be borne
out if tested by acting on them.

We are, however, well over the border line of simple awareness and
into the realm which is usually classified as perception, so let us proceed
to a consideration of that concept.

9. Perceive, Perception. So much has the meaning of this term
changed that one definition has been given as follows: "Perception is
that which comes into consciousness wrhen stimuli, principally light or
sound, impinge on the organism from the outside" [40, p. 250]. Al-
though this seems a bit too general, it does take account of the work of
Hebb, Riesen, and others, which indicates that the impingement of the
stimuli and the meaning given to the stimuli are inseparable parts of a
single experience.

For our own definition we might say that a perception is a hypothesis
or prognosis for action which comes into being in awareness when
stimuli impinge on the organism. When wre perceive "this is a triangle,"
"that is a tree," "this person is my mother,'3 it means that we are making
a prediction that the objects from which the stimuli are received would,
if checked in other ways, exhibit properties we have come to regard,
from our past experience, as being characteristic of triangles, trees,
mother.

Thus we might say that perception and awareness are synonymous,
perception being the narrower term, usually used when we wish to
emphasize the importance of the stimulus in the process, and awareness
the broader term, covering symbolizations and meanings which arise
from such purely internal stimuli as memory traces, visceral changes,
and the like, as well as from external stimuli.

To define perception in this purely psychological fashion is not
meant to deny that it can be defined in physiological fashion by referring
to the impact of a pattern of light rays upon certain nerve cells, for
example. For our purpose, however, the psychological definition seems
more fruitful, and it is in this sense that the term will be used in our
formulations.

10. Subceive, Subception. McCleary and Lazarus [46] formulated
this construct to signify discrimination without awareness. They state
that "even when a subject is unable to report a visual discrimination he
is still able to make a stimulus discrimination at some level below that

200 CARL R. ROGERS

required for conscious recognition.33 Thus it appears that the organism
can discriminate a stimulus and its meaning for the organism without
utilizing the higher nerve centers involved in awareness. It is this capacity
which, in our theory, permits the individual to discriminate an ex-
perience as threatening, without symbolization in awareness of this
threat.

11. Self -experience. This is a term coined by Standal [80], and
defined as being any event or entity in the phenomenal field discriminated
by the individual which is also discriminated as "self," "me," "I," or
related thereto. In general self-experiences are the raw material of which
the organized self-concept is formed.

12. Self, Concept of self, Self -structure. These terms refer to the
organized, consistent conceptual gestalt composed of perceptions of the
characteristics of the "I" or "me" and the perceptions of the relation-
ships of the "I" or "me33 to others and to various aspects of life, to-
gether with the values attached to these perceptions. It is a gestalt
which is available to awareness though not necessarily in awareness.
It is a fluid and changing gestalt, a process, but at any given moment
it is a specific entity which is at least partially definable in operational
terms by means of a Q sort or other instrument or measure. The term
self or self-concept is more likely to be used when we are talking of the
person's view of himself, self -structure when we are looking at this
gestalt from an external frame of reference.

13. Ideal self. Ideal self (or self-ideal) is the term used to denote
the self-concept which the individual would most like to possess, upon
which he places the highest value for himself. In all other respects it is
defined in the same way as the self-concept.

A digression on the case history of a construct. Since the abstrac-
tion which we term the self is one of the central constructs in our
theory, it may be helpful to interpose a somewhat lengthy digression at
this point in our list of definitions in order to relate something of the
development of this construct. In so doing we will also be illustrating
the manner in which most of these defined constructs have come into
being in our theory.

Speaking personally, I began my work with the settled notion that
the "self" was a vague, ambiguous, scientifically meaningless term
which had gone out of the psychologist's vocabulary with the departure
of the introspectionists. Consequently I was slow in recognizing that
when clients were given the opportunity to express their problems and
their attitudes in their own terms, without any guidance or interpreta-

Therapy, Personality^ and Interpersonal Relationships 201

tion, they tended to talk in terms of the self. Characteristic expressions
were attitudes such as these: "I feel I'm not being my real self." "I
wonder who I am? really." i;I wouldn't want anyone to know the real
me.35 "I never had a chance to be myself." "It feels good to let myself
go and just be myself here." "I think if I chip off all the plaster facade
I've got a pretty solid self — a good substantial brick building, under-
neath." It seemed clear from such expressions that the self was an im-
portant element in the experience of the client, and that in some odd
sense his goal was to become his "real self."

Raimy [54] produced a careful and searching definition of the self-
concept which was helpful in our thinking. There seemed to be no
operational way of defining it at that point. Attitudes toward the self
could be measured, however, and Raimy and a number of others began
such research. Self-attitudes were determined, operationally, by the
categorizing of all self-referent terms in interviews preserved in verbatim
form by electrical recording. The categories used had a satisfactory
degree of interjudge reliability, thus making them suitable scientific con-
structs for our work. We were encouraged to find that these self-
referent attitudes altered significantly in therapy as we had hypothesized
they would.

As we focused more upon the concept of the self, clinical experience
again gave us further clues as to its nature. For example, in the process
of change which appeared to occur in therapy, it was not at all un-
common to find violent fluctuation in the concept of the self. A client,
during a given interview, would come to experience himself quite
positively. He felt he was worthwhile, that he could meet life with the
capacities he possessed, and that he was experiencing a quiet confidence.
Three days later he might return with a completely reversed conception
of himself. The same evidence now proved an opposite point. The posi-
tive new choice he had made now was an instance of silly immaturity;
the valid feelings courageously expressed to his colleagues now were
clearly inadequate. Often such a client could date, to the moment, the
point at which, following some very minor incident, the balance was
upset, and his picture of himself had undergone a complete flip-flop.
During the interview it might as suddenly reverse itself again.

Consideration of this phenomenon made it clear that we were not
dealing with an entity of slow accretion, of step-by-step learning, of
thousands of unidirectional conditionings. These might all be involved,
but the product was clearly a gestalt, a configuration in which the alter-
ation of one minor aspect could completely alter the whole pattern. One
was forcibly reminded of the favorite textbook illustration of a gestalt,
the double picture of the old hag and the young woman. Looked at with
one mind set, the picture is clearly that of an ugly old woman. The

202

CARL R. ROGERS

slightest change, and the whole becomes a portrait of an attractive girl.
So with our clients. The self-concept was clearly configurational in
nature.

Our clinical experience gave us another clue to the manner in which
the self functioned. The conventional concept of repression as having to
do with forbidden or socially taboo impulses had been recognized as in-
adequate to fit the facts. Often the most deeply denied impulses and
feelings were positive feelings of love, or tenderness, or confidence in
self. How could one explain the puzzling conglomeration of experience
which seemingly could not be permitted in awareness? Gradually it was
recognized that the important principle was one of consistency with the
self. Experiences which were incongruent with the individual's concept
of himself tended to be denied to awareness, whatever their social char-
acter. We began to see the self as a criterion by which the organism
screened out experiences which could not comfortably be permitted in
consciousness. Lecky's little posthumous book [43] reinforced this line
of thought. We also began to understand other functions of the self in its
regulatory influence on behavior, and the like.

At about this juncture Stephenson's Q technique [81] opened up
the possibility of an operational definition of the self-concept. Im-
mediately, research burgeoned. Though we feel it has barely made a
start in exploiting the possible testing of hypotheses, there have already
been measurements and predictions regarding the self as of this moment,
the self in the past, "myself as I am with my mother," "the self I would
like to be," etc. Probably the most sophisticated and significant of these
studies is that completed by ChodorkofT [10], in which his hypothesis,
stated informally, is as follows: that the greater the agreement between
the individual's self-description and an objective description of him,
the less perceptual defensiveness he will show, and the more adequate
will be his personal adjustment. This hypothesis is upheld and tends to
confirm some important aspects of our theory. In general the various
investigations have agreed in indicating that the self-concept is an im-
portant variable in personality dynamics and that change in the self is
one of the most marked and significant changes occurring in therapy.

It should be recognized that any construct is a more or less arbitrary
abstraction from experience. Thus the self could be defined in many dif-
ferent ways. Hilgard, for example [34], has proposed that it be defined
in such a way as to include unconscious material, not available to aware-
ness, as well as conscious material. Although we recognize that this is
certainly a legitimate way of abstracting from the phenomena, we be-
lieve it is not a useful way because it produces a concept which cannot
at this point be given operational definition. One cannot obtain sufficient
agreement as to the content of the individual's unconscious to make

Therapy, Personality, and Interpersonal Relationships 203

research possible. Hence we believe that it is more fruitful to define the
self-concept as a gestalt which is available to awareness. This has per-
mitted and encouraged a flood of important research.

At all times, however, we endeavor to keep in the forefront of our
thinking the fact that each definition is no more than an abstraction
and that the same phenomena might be abstracted in a different fashion.
One of our group is working on a definition of self which would give
more emphasis to its process nature. Others have felt that a plural
definition, indicating many specific selves in each of various life contexts,
would be more fruitful, and this way of thinking has been embodied in,
for example, XunnaUy's [50] research. So the search continues for a more
adequate conceptualization of this area of our therapeutic experience
and for more adequate technical means of providing operational defini-
tions for the concepts which are formulated.

This concludes our interruption of the list of definitions. It is hoped
that this one example will give an indication of the way in which many
of our basic constructs have developed — not only the self-concept but
the constructs of congruence, incongraence, defensiveness, unconditional
positive regard, locus of evaluation, and the like. Although the process
has been irregular, it has tended to include clinical observation, initial
conceptualization, initial crude research to test some of the hypotheses
involved, further clinical observation, more rigorous formulation of the
construct and its functional relationships, more refined operational
definitions of the construct, more conclusive research.

14. Incongruence between self and experience. In a manner which
will be described in the theory of personality a discrepancy frequently
develops between the self as perceived, and the actual experience of the
organism. Thus the individual may perceive himself as having char-
acteristics a, by and c3 and experiencing feelings x, y\ and z. An accurate
symbolization of his experience would, however, indicate characteristics
c, d} and e, and feelings v, w3 x. When such a discrepancy exists, the
state is one of incongruence between self and experience. This state is
one of tension and internal confusion, since in some respects the in-
dividual's behavior will be regulated by the actualizing tendency, and in
other respects by the self-actualizing tendency, thus producing discordant
or incomprehensible behaviors. What is commonly called neurotic be-
havior is one example, the neurotic behavior being the product of the
actualizing tendency, whereas in other respects the individual is actualiz-
ing the self. Thus the neurotic behavior is incomprehensible to the in-
dividual himself, since it is at variance with what he consciously "wants"
to do, which is to actualize a self no longer congruent with experience.

15. Vulnerability. Vulnerability is the term used to refer to the

204 CARL R. ROGERS

state of incongruence between self and experience, when it is desired
to emphasize the potentialities of this state for creating psychological dis-
organization. When incongruence exists, and the individual is unaware
of it, then he is potentially vulnerable to anxiety, threat, and disorganiza-
tion. If a significant new experience demonstrates the discrepancy so
clearly that it must be consciously perceived, then the individual will be
threatened, and his concept of self disorganized by this contradictory and
unassimilable experience.

16. Anxiety. Anxiety is phenomenologically a state of uneasiness
or tension whose cause is unknown. From an external frame of reference,
anxiety is a state in which the incongruence between the concept of self
and the total experience of the individual is approaching symbolization
in awareness. When experience is obviously discrepant from the self-
concept, a defensive response to threat becomes increasingly difficult.
Anxiety is the response of the organism to the "subception" that such
discrepancy may enter awareness, thus forcing a change in the self-
concept.

17. Threat. Threat is the state which exists when an experience is
perceived or anticipated (subceived) as incongruent with the structure
of the self. It may be regarded as an external view of the same
phenomenon which, from the internal frame of reference, is anxiety.

18. Psychological maladjustment. Psychological maladjustment exists
when the organism denies to awareness, or distorts in awareness, sig-
nificant experiences, which consequently are not accurately symbolized
and organized into the gestalt of the self-structure, thus creating an in-
congruence between self and experience.

It may help to clarify this basic concept of incongruence if we recog-
nize that several of the terms we are defining are simply different
vantage points for viewing this phenomenon. If an individual is in a state
of incongruence between self and experience and we are looking at
him from an external point of view we see him as vulnerable (if he is
unaware of the discrepancy), or threatened (if he has some awareness
of it) . If we are viewing him from a social point of view, then this in-
congruence is psychological maladjustment. If the individual is viewing
himself, he may even see himself as adjusted (if he has no awareness of
the discrepancy) or anxious (if he dimly subceives it) or threatened
or disorganized (if the discrepancy has forced itself upon his awareness).

19. Defense, Defensiveness. Defense is the behavioral response of the
organism to threat, the goal of which is the maintenance of the current
structure of the self. This goal is achieved by the perceptual distortion
of the experience in awareness, in such a way as to reduce the incongruity
between the experience and the structure of the self, or by the denial

Therapy, Personality, and Interpersonal Relationships 205

to awareness of an experience, thus denying any threat to the self. De-
fensiveness is the term denoting a state in which the behaviors are of
the sort described.

20. Distortion in awareness^ Denial to awareness. It is an observed
phenomenon that material which is significantly inconsistent with the
concept of self cannot be directly and freely admitted to awareness. To
explain this the construct of denial or distortion has been developed.
When an experience is dimly perceived (or "subceived" is perhaps the
better term) as being incongraent with the self-structure, the organism
appears to react with a distortion of the meaning of the experience,
(making it consistent with the self) or with a denial of the existence
of the experience, in order to preserve the self-structure from threat. It is
perhaps most vividly illustrated in those occasional moments in therapy
when the therapist's response, correctly heard and understood, would
mean that the client would necessarily perceive openly a serious in-
consistency between his self-concept and a given experience. In such a
case, the client may respond, "I can hear the words you say, and I know
I should understand them, but I just can't make them convey any mean-
ing to me." Here the relationship is too good for the meaning to be
distorted by rationalization, the meaning too threatening to be received.
Hence the organism denies that there is meaning in the communication.
Such outright denial of experience is much less common than the
phenomenon of distortion. Thus if the concept of self includes the
characteristic "I am a poor student" the experience of receiving a high
grade can be easily be distorted to make it congruent with the self by
perceiving in it such meanings as, "That professor is a fool"; "It was
just luck53 ; etc.

21. Intensionality. This term is taken from general semantics. If
the person is reacting or perceiving in an intensional fashion he tends
to see experience in absolute and unconditional terms, to overgeneralize,
to be dominated by concept or belief, to fail to anchor his reactions in
space and time, to confuse fact and evaluation, to rely upon abstractions
rather than upon reality-testing. This term covers the frequently used
concept of rigidity but includes perhaps a wider variety of behaviors
than are generally thought of as constituting rigidity.

It will perhaps be evident that this cluster of definitions all have to
do with the organism's response to threat. Defense is the most general
term: distortion and denial are the mechanisms of defense; intensionality
is a term which covers the characteristics of the behavior of the in-
dividual who is in a defensive state.

22. Congruence, Congruence of self and experience. This is a basic
concept which has grown out of therapeutic experience, in which the

206 CARL R. ROGERS

individual appears to be revising his concept of self to bring it into con-
gruence with his experience, accurately symbolized. Thus he discovers
that one aspect of his experience if accurately symbolized, would be
hatred for his father; another would be strong homosexual desires. He
reorganizes the concept he holds of himself to include these char-
acteristics, which would previously have been inconsistent with self.

Thus when self-experiences are accurately symbolized, and are in-
cluded in the self-concept in this accurately symbolized form, then the
state is one of congruence of self and experience. If this were completely
true of all self-experiences, the individual would be a fully functioning
person, as will be made more clear in the section devoted to this aspect
of our theory. If it is true of some specific aspect of experience, such as
the individual's experience in a given relationship or in a given moment
of time, then we can say that the individual is to this degree in a state
of congruence. Other terms which are in a general way synonymous
are these : integrated, whole, genuine.

23. Openness to experience. When the individual is in no way
threatened, then he is open to his experience. To be open to experience
is the polar opposite of defensiveness. The term may be used in regard to
some area of experience or in regard to the total experience of the
organism. It signifies that every stimulus, whether originating within
the organism or in the environment, is freely relayed through the nervous
system without being distorted or channeled off by any defensive mecha-
nism. There is no need of the mechanism of "subception" whereby the
organism is forewarned of experiences threatening to the self. On the
contrary, whether the stimulus is the impact of a configuration of form,
color, or sound in the environment on the sensory nerves, or a memory
trace from the past, or a visceral sensation of fear, pleasure, or disgust,
it is completely available to the individual's awareness. In the hypo-
thetical person who is completely open to his experience, his concept of
self would be a symbolization in awareness which would be completely
congruent with his experience. There would, therefore, be no possibility
of threat.

24. Psychological adjustment. Optimal psychological adjustment
exists when the concept of the self is such that all experiences are or
may be assimiliated on a symbolic level into the gestalt of the self-
structure. Optimal psychological adjustment is thus synonymous with
complete congruence of self and experience, or complete openness to
experience. On the practical level, improvement in psychological ad-
justment is equivalent to progress toward this end point.

25. Extensionality. This term is taken from general semantics. If
the person is reacting or perceiving in an extensional manner he tends
to see experience in limited, differentiated terms, to be aware of the

Therapy, Personality, and Interpersonal Relationships 207

space-time anchorage of facts, to be dominated by facts, not by concepts,
to evaluate in multiple ways, to be aware of different levels of abstraction,
to test his inferences and abstractions against reality.

26. Mature, Maturity. The individual exhibits mature behavior when
he perceives realistically and in an extensional manner, is not defensive,
accepts the responsibility of being different from others, accepts re-
sponsibility for his own behavior, evaluates experience in terms of the
evidence coming from his own senses, changes his evaluation of ex-
perience only on the basis of new evidence, accepts others as unique in-
dividuals different from himself, prizes himself, and prizes others. (If his
behavior has these characteristics, then there will automatically follow
all the types of behavior which are more popularly thought of as con-
stituting psychological maturity. )

These last five definitions form a cluster which grows out of the con-
cept of congruence. Congruence is the term which defines the state.
Openness to experience is the way an internally congruent individual
meets new experience. Psychological adjustment is congruence as viewed
from a social point of view. Extensional is the term which describes the
specific types of behavior of a congruent individual. Maturity is a
broader term describing the personality characteristics and behavior of
a person who is, in general, congruent.

The concepts in the group of definitions which follow have all been
developed and formulated by Standal [80], and have taken the place
of a number of less satisfactory and less rigorously defined constructs.
Essentially this group has to do with the concept of positive regard, but
since all transactions relative to this construct take place in relationships,
a definition of psychological contact, or minimal relationship, is set down
first.

27. Contact. Two persons are in psychological contact, or have the
minimum essential of a relationship, when each makes a perceived or
subceived difference in the experiential field of the other.

This construct was first given the label of "relationship" but it was
found that this led to much misunderstanding, for it was often under-
stood to represent the depth and quality of a good relationship, or a
therapeutic relationship. The present term has been chosen to signify
more clearly that this is the least or minimum experience which could
be called a relationship. If more than this simple contact between two
persons is intended, then the additional characteristics of that contact
are specified in the theory.

28. Positive regard. If the perception by me of some self-experience
in another makes a positive difference in my experiential field, then I

208 CARL R. ROGERS

am experiencing positive regard for that individual. In general, positive
regard is defined as including such attitudes as warmth, liking, respect,
sympathy, acceptance. To perceive oneself as receiving positive regard is
to experience oneself as making a positive difference in the experiential
field of another.

29. Need for positive regard. It is postulated by Standal that a basic
need for positive regard, as defined above, is a secondary or learned
need, commonly developed in early infancy. Some writers have looked
upon the infant's need for love and affection as an inherent or instinctive
need. Standal is probably on safer ground in regarding it as a learned
need. By terming it the need for positive regard, he has, it is believed,
selected out the significant psychological variable from the broader terms
usually used.

30. Unconditional positive regard. Here is one of the key constructs
of the theory, which may be defined in these terms: if the self-experiences
of another are perceived by me in such a way that no self-experience can
be discriminated as more or less worthy of positive regard than any other,
then I am experiencing unconditional positive regard for this individual.
To perceive oneself as receiving unconditional positive regard is to per-
ceive that of one's self-experiences none can be discriminated by the
other individual as more or less worthy of positive regard.

Putting this in simpler terms, to feel unconditional positive regard
toward another is to^prize'' him (to use Dewey's term, recently used
in this sense by Butler) . This means to value the person, irrespective of
the differential values which one might place on his specific behaviors.
A parent "prizes" his child, though he may not value equally all of his
behaviors. Acceptance is another term which has been frequently used
to convey this meaning, but it perhaps carries more misleading con-
notations than the phrase which Standal has coined. In general, how-
ever, acceptance and prizing are synonymous with unconditional positive
regard.

This construct has been developed out of the experiences of therapy,
where it appears that one of the potent elements in the relationship is
that the therapist "prizes" the whole person of the client. It is the fact
that he feels and shows an unconditional positive regard toward the ex-
periences of which the client is frightened or ashamed, as well as to-
ward the experiences with which the client is pleased or satisfied, that
seems effective in bringing about change. Gradually the client can feel
more acceptance of all of his own experiences, and this makes him again
more of a whole or congruent person, able to function effectively. This
clinical explanation will, it is hoped, help to illuminate the meaning
contained in the rigorous definition.

31. Regard complex. The regard complex is a construct defined by

Therapy, Personality, and Interpersonal Relationships 209

Standal as all those self-experiences, together with their interrelation-
ships, which the individual discriminates as being related to the positive
regard of a particular social other.

This construct is intended to emphasize the gestalt nature of trans-
actions involving positive or negative regard, and their potency. Thus,
for example, if a parent shows positive regard to a child in relationship
to a specific behavior, this tends to strengthen the whole pattern of
positive regard which has previously been experienced as coming from
that parent. Likewise specific negative regard from this parent tends to
weaken the whole configuration of positive regard.

32. Positive self -regard. This term is used to denote a positive regard
satisfaction which has become associated with a particular self-experience
or a group of self-experiences, in which this satisfaction is independent
of positive regard transactions with social others. Though it appears that
positive regard must first be experienced from others, this results in a
positive attitude toward self which is no longer directly dependent on
the attitudes of others. The individual, in effect, becomes his own
significant social other.

33. Need for self -regard. It is postulated that a need for positive
self-regard is a secondary or learned need, related to the satisfaction of
the need for positive regard by others.

34. Unconditional self-regard. When the individual perceives him-
self in such a way that no self-experience can be discriminated as more or
less worthy of positive regard than any other, then he is experiencing
unconditional positive self-regard.

35. Conditions of worth. The self-structure is characterized by a
condition of worth when a self-experience or set of related self-ex-
periences is either avoided or sought solely because the individual dis-
criminates it as being less or more worthy of self-regard.

This important construct has been developed by Standal to take the
place of "introjected value," which was a less exact concept used in
earlier formulations. A condition of worth arises when the positive
regard of a significant other is conditional, when the individual feels
that in some respects he is prized and in others not. Gradually this same
attitude is assimilated into his own self-regard complex, and he values
an experience positively or negatively solely because of these conditions
of worth which he has taken over from others, not because the ex-
perience enhances or fails to enhance his organism.

It is this last phrase which deserves special note. When the individual
has experienced unconditional positive regard, then a new experience
is valued or not, depending on its effectiveness in maintaining or en-

210 CARL R. ROGERS

hancing the organism. But if a value is "introjected" from a significant
other, then this condition of worth is applied to an experience quite
without reference to the extent to which it maintains or enhances the
organism. It is an important specific instance of inaccurate symboliza-
tion, the individual valuing an experience positively or negatively, as if
in relation to the criterion of the actualizing tendency, but not actually
in relation to it. An experience may be perceived as organismically
satisfying, when in fact this is not true. Thus a condition of worth, be-
cause it disturbs the valuing process, prevents the individual from
functioning freely and with maximum effectiveness.

36. Locus of evaluation. This term is used to indicate the source
of evidence as to values. Thus an internal locus of evaluation, within the
individual himself, means that he is the center of the valuing process,
the evidence being supplied by his own senses. When the locus, of
evaluation resides in others, their judgment as to the value of an object
or experience becomes the criterion of value for the individual.

37. Organismic valuing process. This concept describes an ongoing-
process in which values are never fixed or rigid, but experiences are
being accurately symbolized and continually and freshly valued in terms
of the satisfactions organismically experienced; the organism experiences
satisfaction in those stimuli or behaviors which maintain and enhance
the organism and the self, both in the immediate present and in the long
range. The actualizing tendency is thus the criterion. The simplest ex-
ample is the infant who at one moment values food, and when satiated,
is disgusted with it; at one moment values stimulation, and soon after,
values only rest; who finds satisfying that diet which in the long run most
enhances his development.

38. Internal frame of reference. This is all of the realm of experience
which is available to the awareness of the individual at a given moment.
It includes the full range of sensations, perceptions, meanings, and
memories, which are available to consciousness.

The internal frame of reference is the subjective world of the in-
dividual. Only he knows it fully. It can never be known to another ex-
cept through empathic inference and then can never be perfectly known.

39. Empathy. The state of empathy, or being empathic, is to per-
ceive the internal frame of reference of another with accuracy, and with
the emotional components and meanings which pertain thereto, as if one
were the other person, but without ever losing the ccas if condition.
Thus it means to sense the hurt or the pleasure of another as he senses
it, and to perceive the causes thereof as he perceives them, but without

Therapy, Personality, and Interpersonal Relationships 211

ever losing the recognition that it is as if I were hurt or pleased, etc.
If this "as if" quality is lost, then the state is one of Identification.

40. External frame of reference. To perceive solely from one's own
subjective internal frame of reference without empathizing with the ob-
served person or object, is to perceive from an external frame of
reference. The "empty organism" school of thought in psychology is an
example of this. Thus the observer says that an animal has been
stimulated when the animal has been exposed to a condition which, in
the observer's subjective frame of reference, is a stimulus. There is no
attempt to understand, empathically, whether this is a stimulus in the
animal's experiential field. Likewise the observer reports that the animal
emits a response when a phenomenon occurs which, in the observer's
subjective field, is a response.

We generally regard all "objects" (stones, trees, or abstractions) from
this external frame of reference since we assume that they have no "ex-
perience" with which we can empathize. The other side of* this coin is
that anything perceived from an external frame of reference (whether
an inanimate thing, an animal, or a person) becomes for us an "object"
because no empathic inferences are made.

This cluster of three ways of knowing deserves some further com-
ment. In so far as we are considering knowledge of human beings we
might say that these ways of knowing exist on a continuum. They range
from one's own complete subjectivity in one's own internal frame of
reference to one's own complete subjectivity about another (the external
frame of reference). In between lies the range of empathic inference
regarding the subjective field of another.

Each of these ways of knowing is essentially a formulation of hypoth-
eses. The differences lie in the way the hypotheses are checked. In my
own internal frame of reference if I experience love or hate, enjoyment
or dislike, interest or boredom, belief or disbelief, the only way I can
check these hypotheses of experience is by further focusing on my
experience. Do I really love him? Am I really enjoying this? Do I really
believe this? are questions which can only be answered by checking
with my own organism. (If I try to find out whether I really love him
by checking with others, then I am observing myself as an object, am
viewing myself from an external frame of reference. )

Although in the last analysis each individual lives in and by his
own subjective knowledge, this is not regarded socially as "knowledge"
and certainly not as scientific knowledge.

Knowledge which has any "certainty," in the social sense, involves
the use of empathic inference as a means of checking, but the direction
of that empathy differs. When the experience of empathic understanding
is used as a source of knowledge, one checks one's empathic inferences

212 CARL R. ROGERS

with the subject, thus verifying or disproving the inferences and hypoth-
eses implicit in such empathy. It is this way of knowing which we have
found so fruitful in therapy. Utilizing empathic inference to the fullest,
the knowledge thus gained of the client's subjective world has led to
understanding the basis of his behavior and the process of personality
change.

In knowing a person or an object from the external frame of refer-
ence, our implicit hypotheses are checked with other people, but not with
the subject of our concern. Thus a rigorous behaviorist believes that S
is a stimulus for his experimental animal and R is a response, because
his colleagues and even the man in the street agree with him and re-
gard S and R in the same way. His empathic inferences are made
in regard to the internal frame of reference of his colleagues, rather than
in regard to the internal frame of reference of the animal.

Science involves taking an external frame of reference, in which we
check our hypotheses basically through empathic inferences as to the
internal frame of reference of our colleagues. They perform the same
operations we have (either actually or through symbolic representation),
and if they perceive the same events and meanings, then we regard our
hypotheses as confirmed.

The reason for thus elaborating the different ways of knowing is that
it seems to us that all ways of knowing have their usefulness, and that
confusion arises only when one is not clear as to the type of knowledge
which is being specified. Thus in the theory of therapy which follows one
will find certain conditions of therapy specified as subjective experiencing
states, another as an empathic knowledge of the client, and yet the
scientific checking of the hypotheses of the theory can only be done from
an external frame of reference.

I. A THEORY OF THERAPY AND PERSONALITY CHANGE

This theory is of the if-then variety. If certain conditions exist
(independent variables), then a process (dependent variable) will
occur which includes certain characteristic elements. If this process
(now the independent variable) occurs, then certain personality and
behavioral changes (dependent variables) will occur. This will be made
specific.

In this and the following sections the formal statement of the
theory is given briefly, in smaller type. The italicized terms or
phrases in these formal statements have been defined in the previous
section and are to be understood as defined. The remaining paragraphs
are explanatory and do not follow the rigorous pattern of the formal
statements.

Therapy, Personality, and Interpersonal Relationships 213

A. Conditions of the Therapeutic Process

For therapy to occur it is necessary that these conditions exist.

1. That two persons are in contact.

2. That the first person, whom we shall term the client, is in a state of
incongruence, being vulnerable, or anxious.

3. That the second person, whom we shall term the therapist, is con-
gruent in the relationship.

4. That the therapist is experiencing unconditional positive regard to-
ward the client.

5. That the therapist is experiencing an empathic understanding of the
client's internal frame of reference.

6. That the client perceives, at least to a minimal degree, conditions 4
and 53 the unconditional positive regard of the therapist for him, and the
empathic understanding of the therapist.

Comment. These seem to be the necessary conditions of therapy,
though other elements are often or usually present. The process is more
likely to get under way if the client is anxious, rather than merely
vulnerable. Often it is necessary for the contact or relationship to be of
some duration before the therapeutic process begins. Usually the em-
pathic understanding is to some degree expressed verbally, as well as
experienced. But the process often commences with only these minimal
conditions, and it is hypothesized that it never commences without these
conditions being met.

The point which is most likely to be misunderstood is the omission
of any statement that the therapist communicates his empathic under-
standing and his unconditional positive regard to the client. Such a state-
ment has been omitted only after much consideration, for these reasons.
It is not enough for the therapist to communicate, since the communica-
tion must be received, as pointed out in condition 6, to be effective.
It is not essential that the therapist intend such communication, since
often it is by some casual remark, or involuntary facial expression, that
the communication is actually achieved. However, if one wishes to stress
the communicative aspect which is certainly a vital part of the living
experience, then condition 6 might be worded in this fashion :

6. That the communication to the client of the therapist's empathic
understanding and unconditional positive regard is, at least to a minimal
degree, achieved.

The element which will be most surprising to conventional therapists
is that the same conditions are regarded as sufficient for therapy, regard-
less of the particular characteristics of the client. It has been our ex-
perience to date that although the therapeutic relationship is used dif-

214 CARL R. ROGERS

ferently by different clients, it is not necessary nor helpful to manipulate
the relationship in specific ways for specific kinds of clients. To do this
damages, it seems to us, the most helpful and significant aspect of the
experience, that it is a genuine relationship between two persons, each
of whom is endeavoring, to the best of his ability, to be himself in the
interaction.1

The "growing edge" of this portion of the theory has to do with
point 3, the congruence or genuineness of the therapist in the relation-
ship. This means that the therapist's symbolization of his own ex-
perience in the relationship must be accurate, if therapy is to be most
effective. Thus if he is experiencing threat and discomfort in the
relationship, and is aware only of an acceptance and understanding, then
he is not congruent in the relationship and therapy will suffer. It seems
important that he should accurately "be himself3 in the relationship,
whatever the self of that moment may be.

Should he also express or communicate to the client the accurate
symbolization of his own experience? The answer to this question is still
in an uncertain state. At present we would say that such feelings should
be expressed, if the therapist finds himself persistently focused on his own
feelings rather than those of the client, thus greatly reducing or eliminat-
ing any experience of empathic understanding, or if he finds himself
persistently experiencing some feeling other than unconditional positive
regard. To know whether this answer is correct demands further testing
of the hypothesis it contains, and this is not simple since the courage
to do this is often lacking, even in experienced therapists. When the thera-
pist's real feelings are of this order: "I find myself fearful that you are
slipping into a psychosis," or CCI find myself frightened because you are
touching on feelings I have never been able to resolve," then it is difficult
to test the hypothesis, for it is very difficult for the therapist to express
such feelings.

Another question which arises is this : is it the congruence, the whole-
ness, the integration of the therapist in the relationship which is im-
portant, or are the specific attitudes of empathic understanding and un-

1This paragraph may have to be rewritten if a recent study of Klrtncr [42]
is confirmed. Kirtner has found, in a group of 26 cases from the Counseling
Center at the University of Chicago, that there are sharp differences in the client's
mode of approach to the resolution of life difficulties and that these differences
are related to success in therapy. Briefly, the client who sees his problem as in-
volving his relationships, and who feels that he contributes to this problem and
wants to change it, is likely to be successful. The client who externalizes his
problem and feels little self-responsibility is much more likely to be a failure.
Thus the implication is that different conditions of therapy may be necessary to
make personality change possible in this latter group. If this is verified, then the
theory will have to be revised accordingly.

Therapy, Personality, and Interpersonal Relationships 215

conditional positive regard vital? Again the final answer is unknown, but
a conservative answer, the one we have embodied in the theory, is that
for therapy to occur the wholeness of the therapist in the relationship is
primary, but a part of the congruence of the therapist must be the ex-
perience of unconditional positive regard and the experience of empathic
understanding.

Another point worth noting is that the stress is upon the experience
in the relationship. It is not to be expected that the therapist is a com-
pletely congruent person at all times. Indeed if this were a necessary
condition there would be no therapy. But it is enough if in this particular
moment of this immediate relationship with this specific person he is
completely and fully himself, with his experience of the moment being
accurately symbolized and integrated into the picture he holds of himself.
Thus it is that imperfect human beings can be of therapeutic assistance
to other imperfect human beings.

The greatest flaw in the statement of these conditions is that they
are stated as if they were all-or-none elements, whereas conditions 2 to 6
all exist on continua. At some later date we may be able to say that
the therapist must be genuine or congruent to such and such a degree
in the relationship, and similarly for the other items. At the present we
can only point out that the more marked the presence of conditions 2 to
6, the more certain it is that the process of therapy will get under way,
and the greater the degree of reorganization which will take place. This
function can only be stated qualitatively at the present time.

Evidence. Confirmatory evidence, particularly of item 5, is found
in the studies by Fiedler [19, 20] and Quinn [52]. Fiedler's study showed
that experienced therapists of different orientations created relationships
in which one of the most prominent characteristics was the ability to
understand the client's communications with the meaning these com-
munications had for the client. Quinn found that the quality of therapist
communication was of crucial significance in therapy. These studies add
weight to the importance of empathic understanding.

Seeman [75] found that increase in the counselor's liking for the
client during therapy was significantly associated with therapeutic success.
Both Seeman and Lipkin [44] found that clients who felt themselves to
be liked by the therapist tended to be more successful. These studies
tend to confirm condition 4 (unconditional positive regard) and condi-
tion 6 (perception of this by the client) .

Though clinical experience would support condition 2, the client's
vulnerability or anxiety, there is little research which has been done in
terms of these constructs. The study by Gallagher [21] indicates that
less anxious clients tend never to become involved in therapy, but drop
out.

216 CARL R. ROGERS

B. The Process of Therapy

When the preceding conditions exist and continue, a process is set in
motion which has these characteristic directions :

1. The client is increasingly free in expressing his feelings., through
verbal and/or motor channels.

2. His expressed feelings increasingly have reference to the self, rather
than nonself .

3. He increasingly differentiates and discriminates the objects of his
feelings and perceptions., including his environment, other persons, his self,
his experiences, and the interrelationships of these. He becomes less in-
tenslonal and more extensional in his perceptions, or to put it in other terms,
his experiences are more accurately symbolized.

4. His expressed feelings increasingly have reference to the incongruity
between certain of his experiences and his concept of self.

5. He comes to experience in awareness the threat of such incongruence.
a. This experience of threat is possible only because of the continued

unconditional positive regard of the therapist, which is extended to
incongruence as much as to congruence, to anxiety as much as to
absence of anxiety.

6. He experiences fully, in awareness, feelings which have in the past
been denied to awareness, or distorted in awareness.

7. His concept of self becomes reorganized to assimilate and include
these experiences which have previously been distorted in or denied to
awareness.

8. As this reorganization of the self-structure continues, his concept of
self becomes increasingly congruent with his experience; the self now in-
cluding experiences which previously would have been too threatening to be
in awareness.

a. A corollary tendency is toward fewer perceptual distortions in aware-
ness, or denials to awareness, since there are fewer experiences which
can be threatening. In other words, defensiveness is decreased.

9. He becomes increasingly able to experience, without a feeling of
threat, the therapist's unconditional positive regard.

10. He increasingly feels an unconditional positive self-regard.

11. He increasingly experiences himself as the locus of evaluation.

12. He reacts to experience less in terms of his conditions of worth and
more in terms of an organismlc valuing process.

Comment. It cannot be stated with certainty that all of these are
necessary elements of the process, though they are all characteristic. Both
from the point of view of experience, and the logic of the theory, 3, 6, 73

Therapy, Personality, and Interpersonal Relationships 217

8, 10, 12, are necessary elements in the process. Item 5a is not a logical
step in the theory but is put in as an explanatory note.

The element which will doubtless be most puzzling to the reader
is the absence of explanatory mechanisms. It may be well to restate our
scientific purpose in terms of an example. // one strokes a piece of steel
with a magnet, and if one places the piece of steel so that it can rotate
freely, then it will point to the north. This statement of the if-then
variety has been proved thousands of times. Why does it happen? There
have been various theoretical answers, and one would hesitate to say,
even now, that we know with certitude why this occurs.

In the same way I have been saying in regard to therapy, "If these
conditions exist, then these subsequent events will occur." Of course we
have speculations as to why this relationship appears to exist, and those
speculations will be increasingly spelled out as the presentation continues.
Nevertheless the most basic element of our theory is that if the described
conditions exist, then the process of therapy occurs, and the events
which are called outcomes will be observed. We may be quite wrong
as to why this sequence occurs. I believe there is an increasing body of
evidence to show that it does occur.

Evidence. There is confirming evidence of varying degrees of
relevance for a number of these items describing the therapeutic process.
Item 2 (increasing self-reference) is supported by our many recorded
therapeutic cases, but has not been reduced to a statistical finding.
Stock's study [82] supports item 3, indicating that client self-referent
expressions become more objective, less strongly emotional. Mitchell [47]
shows that clients become more extensional.

Objective clinical evidence supporting items 4, 5, and 6 is provided
in the form of recordings from a case by Rogers [67].

The findings of Vargas [85] are relevant to item 7, indicating the
way the self is reorganized in terms of emergent new self-perceptions.
Hogan [36] and Haigh [29] have studied the decrease in defensiveness
during the process, as described in item 8a, their findings being con-
firmatory. The increased congruence of self and experience is supported
in an exhaustive single case investigation by Rogers [67]. That such
congruence is associated with lack of defensiveness is found by Chodor-
koflf [10].

Item 10, the increase in the client's positive self-regard, is well
attested by the studies of Snyder [79], Seeman [76], Raimy [55], Stock
[82], Strom [83], Sheerer [78], Lipkin [44]. The client's trend toward
experiencing himself as the locus of evaluation is most clearly shown
by Raskin's research [56], but this is supported by evidence from Sheerer
[78], Lipkin [44], Kessler [41].

218 CARL R. ROGERS

C. Outcomes in Personality and Behavior

There is no clear distinction between process and outcome. Items of
process are simply differentiated aspects of outcome. Hence the statements
which follow could have been included under process. For reasons of con-
venience in understanding, there have been grouped here those changes
which are customarily associated with the terms outcomes, or results, or are
observed outside of the therapeutic relationship. These are the changes
which are hypothesized as being relatively permanent :

1. The client is more congruent, more open to his experience, less
defensive.

2. He is consequently more realistic, objective, extensional in his per-
ceptions.

3. He is consequently more effective in problem solving,

4. His psychological adjustment is improved, being closer to the
optimum.

a. This is owing to, and is a continuation of, the changes in self-structure
described in B7 and BB.

5. As a result of the increased congruence of self and experience (C4
above) his vulnerability to threat is reduced.

6. As a consequence of C2 above, his perception of his ideal self is
more realistic, more achievable.

7. As a consequence of the changes in C4 and C5 his self is more con-
gruent with his ideal self.

8. As a consequence of the increased congruence of self and ideal self
(C6) and the greater congruence of self and experience, tension of all types
is reduced — physiological tension, psychological tension, and the specific
type of psychological tension defined as anxiety.

9. He has an increased degree of positive self-regard.

10. He perceives the locus of evaluation and the locus of choice as
residing within himself.

a. As a consequence of C9 and CIO he feels more confident and more
self-directing.

b. As a consequence of Cl and CIO, his values are determined by an
organismic valuing process.

11. As a consequence of C13 and C2, he perceives others more realisti-
cally and accurately.

12. He experiences more acceptance of others, as a consequence of less
need for distortion of his perceptions of them.

13. His behavior changes in various ways.

a. Since the proportion of experience assimilated into the self -structure
is increased, the proportion of behaviors which can be "owned" as
belonging to the self is increased.

Therapy, Personality, and Interpersonal Relationships 219

b. Conversely, the proportion of behaviors which are disowned as self-
experiences, felt to be "not myself/' is decreased.

c. Hence his behavior is perceived as being more within his control.

14. His behavior is perceived by others as more socialized, more mature.

15. As a consequence of Cl, 2, 3, his behavior is more creative, more
uniquely adaptive to each new situation, and each new problem3 more fully
expressive of his own purposes and values.

Comment. The statement in part C which is essential is statement Cl.
Items 2 through 15 are actually a more explicit spelling out of the
theoretical implications of statement 1. The only reason for including
them is that though such implications follow readily enough from the
logic of the theory, they are often not perceived unless they are pointed
out.

Evidence. There is much confirmatory and some ambiguous or non-
confirming evidence of the theoretical statement of outcomes. Grummon
and John [28] find a decrease in defensiveness, basing judgements on
the TAT. Hogan [36] and Haigh [29] also supply some scanty evidence
on this point. As to the greater extensionality of perceptions (item 2),
Jonietz [38] finds that therapy produces changes in perceptions and
Mitchell [47] finds these changes to be in the direction of extensionality.

Item 4, stating that adjustment is improved, is supported by evidence
based upon TAT, Rorschach, counselor rating, and other indexes, in the
studies of Dymond [15, 16], Grummon and John [28], Haimowitz [30],
Muench [49], Mosak [48], Cowen and Combs [13]. Carr [8], however,
found no evidence of change in the Rorschach in nine cases.

Rudikoff [73] found that the self-ideal becomes more achievable,
as stated in item 6. The increased congruence of self and ideal has been
confirmed by Butler and Haigh [7], Hartley [33], and its significance for
adjustment supported by Hanlon, Hofstaetter, and O'Connor (32).

The decrease in physiological tension over therapy is attested by
the studies of Thetford [84] and Anderson [1]. The reduction in psy-
chological tension as evidenced by the Discomfort-Relief Quotient has
been confirmed by many investigators : Assum and Levy [4] , Gofer and
Chance [12], Kaufman and Raimy [39], N. Rogers [72], Zimmerman
[36].

The increase in positive self-regard is well attested, as indicated in
IB, Evidence. The shift in the locus of evaluation and choice is supported
in the evidence provided by Raskin [56] and Sheerer [78]. Rudikoff [73]
presents evidence which suggests that others may be perceived with
greater realism. Sheerer [78] and Stock [82] and Rudikoff [73] show
that others are perceived in a more acceptant fashion as postulated
in item 11. Gordon and Cartwright [25] provide evidence which is

220 CARL R. ROGERS

complex but In general nonconfirming on this point. M. Haimowitz
[30] also has findings which seem to indicate that nonacceptance of
minority groups may be more openly expressed.

The behavior changes specified in items 13 and 14 find support in
the Rogers study [68] showing that in improved cases both the client
and his friends observe greater maturity in his behavior. Hoffman [35]
finds that the behavior the client describes in the interviews becomes
more mature. Jonietz's study of [38] of perception of ink blots might lend
some support to the postulate of item 15.

Comments on the theory of therapy. It is to be noted that this
theory of therapy involves, basically, no intervening variables. The condi-
tions of therapy, given in A, are all operationally definable, and some
have already been given rather crude operational definitions in research
already conducted. The theory states that if A exists, then B and C will
follow. B and C are measurable events, predicted by A.

It should also be pointed out that the logic of the theory is such that :
if A, then B; if A, then B and C; if A, then C (omitting conisderation of
5), if B., then C (omitting consideration of A) .

Specification of functional relationships. At this point, the functional
relationships can only be stated in general and qualitative form. The
greater the degree of the conditions specified in A} the more marked
or more extensive will be the process changes in B, and the greater or
more extensive the outcome changes specified in C. Putting this in more
general terms, the greater the degree of anxiety in the client, congruence
in the therapist in the relationship, acceptance and empathy experienced
by the therapist, and recognition by the client of these elements, the
deeper will be the process of therapy, and the greater the extent of
personality and behavioral change. To revert now to the theoretical
logic, all we can say at present is that

B = (f)A C=(f)A B + C = (f)A C = (f)B

Obviously there are many functional interrelationships not yet
specified by the theory. For example, if anxiety is high, is congruence on
the part of the therapist less necessary? There is much work to be done
in investigating the functional relationships more fully.

D. Some Conclusions Regarding the Nature of the Individual

From the theory of therapy as stated above, certain conclusions are
implicit regarding the nature of man. To make them explicit involves little
more than looking at the same hypotheses from a somewhat different
vantage point. It is well to state them explicitly, however, since they con-
stitute an important explanatory link of a kind which gives this theory what-

Therapy, Personality, and Interpersonal Relationships 221

ever uniqueness it may possess. They also constitute the impelling reason for
developing a theory of personality. If the individual is what he is revealed
to be in therapy, then what theory would account for such an individual?
We present these conclusions about the characteristics of the human
organism :

1. The individual possesses the capacity to experience in awareness the
factors in his psychological maladjustment, namely, the incongruences be-
tween his self-concept and the totality of his experience.

2. The individual possesses the capacity and has the tendency to re-
organize his self-concept in such a way as to make it more congruent with
the totality of his experience, thus moving himself away from a state of
psychological maladjustment, and toward a state of psychological adjust-
ment.

3. These capacities and this tendency, when latent rather than evident,
will be released in any interpersonal relationship in which the other person
is congruent in the relationship, experiences unconditional positive regard
toward, and em pat hie understanding of the individual, and achieves some
communication of these attitudes to the individual. (These are, of course,
the characteristics already given under 1,43, 4, 5, 6.)

It is this tendency which, in the following theory of personality, is
elaborated into the tendency toward actualization.

I believe it is obvious that the basic capacity which is hypothesized
is of very decided importance in its psychological and philosophical im-
plications. It means that psychotherapy is the releasing of an already
existing capacity in a potentially competent individual, not the expert
manipulation of a more or less passive personality.2 Philosophically it
means that the individual has the capacity to guide, regulate, and control
himself, providing only that certain definable conditions exist. Only in
the absence of these conditions, and not in any basic sense, is it necessary
to provide external control and regulation of the individual.

II. A THEORY OF PERSONALITY

In endeavoring to order our perceptions of the individual as he
appears in therapy, a theory of the development of personality, and of the
dynamics of behavior, has been constructed. It may "be well to repeat
the warning previously given, and to note that the initial propositions

2 In order to correct a common misapprehension it should be stated that this
tentative conclusion in regard to human capacity grew out of continuing work
with clients in therapy. It was not an assumption or bias with which we started
our therapeutic endeavors. A brief personal account of the way in which this
conclusion was forced upon me is contained in an autobiographical paper [69],

222 CARL R. ROGERS

of this theory are those which are furthest from the matrix of our ex-
perience and hence are most suspect. As one reads on, the propositions
become steadily closer to the experience of therapy. As before, the defined
terms and constructs are italicized, and are to be understood as previously
defined.

A. Postulated Characteristics of the Human Infant

It is postulated that the individual, during the period of infancy, has at
least these attributes.

1. He perceives his experience as reality. His experience is his reality.
a. As a consequence he has greater potential awareness of what reality
is for him than does anyone else, since no one else can completely
assume his internal frame of referenced- ' ^
*> 2. He has an inherent tendency toward actualizing his organism.

3. He interacts with his reality in terms of his basic actualizing tendency.
Thus his behavior is the goal-directed attempt of the organism to satisfy
the experienced needs for actualization in the reality as perceived.
'?. 4. In this interaction he behaves as an organized whole, as a gestalt.

5. He engages in an organismic valuing process, valuing experience with
reference to the actualizing tendency as a criterion. Experiences which are
perceived as maintaining or enhancing the organism are"valued positively.
Those which are perceived as negating such maintenance or enhancement
are valued negatively.

6. He behaves with adience toward positively valued experiences and
with avoidance toward those negatively valued.

Comment. In this view as formally stated, the human infant is seen
as having an inherent motivational system (which he shares in common
with all living things) and a regulatory system (the valuing process)
which by its "feedback" keeps the organism "on the beam33 of satis-
fying his motivational needs. He lives in an environment which for the-
oretical purposes may be said to exist only in him, or to be of his own
creation.

This last point seems difficult for some people to comprehend. It is
the perception of the environment which constitutes the environment,
regardless as to how this relates to some "real33 reality which we may
philosophically postulate. The infant may be picked up by a friendly,
affectionate person. If his perception of the situation is that this is a
strange and frightening experience, it is this perception, not the "reality"
or the "stimulus" which will regulate his behavior. To be sure, the rela-
tionship with the environment is a transactional one, and if his con-
tinuing experience contradicts his initial perception, then in time his

Therapy, Personality, and Interpersonal Relationships 223

perception will change. But the effective reality which influences behavior
is at all times the perceived reality. We can operate theoretically from
this base without having to resolve the difficult question of what "really"
constitutes reality.

Another comment which may be in order is that no attempt has
been made to supply a complete catalogue of the equipment with which
the infant faces the world. Whether he possesses instincts, or an in-
nate sucking reflex, or an innate need for affection, are interesting
questions to pursue, but the answers seem peripheral rather than essential
to a theory of personality.

B. The Development of the Self

1. In line with the tendency toward differentiation which is a part of
the actualizing tendency, a portion of the individual's experience becomes
differentiated and symbolized in an .awareness of being, awareness of func-
tioning. Such awareness may be described as self-experience.

2. This representation in awareness of being and functioning, becomes
elaborated, through interaction with the environment, particularly the en-
vironment composed of significant others, into a concept of self, a perceptual
object in his experiential field.

Comment. These are the logical first steps in the development of the
self. It is by no means the way the construct developed in our own
thinking, as has been indicated in the section of definitions. (A digression
on the case history of a construct, p. 200. )

C. The Need for Positive Regard

1. As the awareness of self emerges, the individual develops a need for
positive regard. This need is universal in human beings, and in the in-
dividual, is pervasive and persistent. Whether it is an inherent or learned
need is irrelevant to the theory. Standal [80], who formulated the concept,
regards it as the latter.

a. The satisfaction of this need is necessarily based upon inferences re-
garding the experiential field of another.

( 1 ) Consequently it is often ambiguous.

b. It is associated with a/ very wide range of the individual's experiences.

c. It is reciprocal, in that when an individual discriminates himself as
satisfying another's need for positive regard, he necessarily experiences
satisfaction of his own need for positive regard.

( 1 ) Hence it is rewarding both to satisfy this need in another, and to
experience the satisfaction of one's own need by another.

224 CARL R. ROGERS

d. It is potent,, in that the positive regard of any social other is com-
municated to the total regard complex which the individual associates
with that social other.

(1) Consequently the expression of positive regard by a significant
social other can become more compelling than the organismic
valuing process, and the individual becomes more adient to
the positive regard of such others than toward experiences which
are of positive value in actualizing the organism.

D. The Development of the Need for Self-regard

1. The positive regard satisfactions or frustrations associated with any
particular self-experience or group of self-experiences come to be experienced
by the individual independently of positive regard transactions with social
others. Positive regard experienced in this fashion is termed self-regard.

2. A need for self-regard develops as a learned need developing out of
the association of self-experiences with the satisfaction or frustration of the
need for positive regard. —»«•-•

3. The individual thus comes to experience positive regard or loss of
positive regard independently of transactions with any social other. He
becomes in a sense his own significant social other.

4. Like positive regard, self-regard which is experienced in relation to
any particular self-experience or group of self-experiences, is communicated
to the total self-regard complex.

E. The Development of Conditions of Worth

1. When self -experiences of the individual are discriminated by sig-
nificant others as being more or less worthy of positive regard, then self-
regard becomes similarly selective.

2. When a self-experience is avoided (or sought) solely because it is
less (or more) worthy of self -regard, the individual is said to have acquired
a condition of worth.

3. If an individual should experience only unconditional positive regard,
then no conditions of worth would develop, self-regard would be uncondi-
tional, the needs for positive regard and self-regard would never be at
variance with jgrfamsmic^n^ the individual would continue to
be psychologically adjusted., and would be fully functioning. This chain of
events is hypothetically possible, and hence important theoretically, though
it does not appear to occur in actuality.

Comment. This is an important sequence in personality development,
stated more fully by Standal [80]. It may help to restate the sequence in
informal, illustrative, and much less exact terms.

Therapy, Personality, and Interpersonal Relationships 225

The infant learns to need love. Love is very satisfying, but to know
whether he is receiving it or not he must observe his mother's face,
gestures, and other ambiguous signs. He develops a total gestalt as to the
way he is regarded by his mother and each new experience of love or
rejection tends to alter the whole gestalt. Consequently each behavior
on his mother's part such as a specific disapproval of a specific behavior
tends to be experienced as disapproval in general. So important is this
to the infant that he comes to be guided in his behavior not by the degree
to which an experience maintains or enhances the organism, but by the
likelihood of receiving maternal love.

Soon he learns to view himself in much the same way, liking or
disliking himself as a total configuration. He tends, quite independently
of his mother or atib^i^_to_yiew himseli andjns behavior in the same
way they have. This means that some behaviors are regarded positively
which are not actually experienced organismically as satisfying. Other
behaviors are regarded negatively which are not actually experienced
as unsatisfying. It is when he behaves in accordance with these intro-
jected values that he may be said to have acquired conditions_of_worth.
He cannot regard himself positively, as having worth, unless he lives in
terms of these conditions. He now reacts with adience or avoidance
toward certain behaviors solely because of these introjected conditions
of self-regard, quite without reference to the organismic consequences
of these behaviors. This is what is meant by living in terms of intro-
jected values (the phrase formerly used) or conditions of worth.

It is not theoretically necessary that such a sequence develop. If the
infant always felt prized, if his own feelings were always accepted even
though some behaviors were inhibited, then no conditions of worth
would develop. This could at least theoretically be achieved if the
parental attitude was genuinely of this sort: "I can understand how satis-
fying it feels to you to hit your baby brother (or to defecate when and
where you please, or to destroy things) and I love you and am quite
willing for you to have those feelings. But I am quite willing for me to
have my feelings, too, and I feel very distressed when your brother is
hurt, (or annoyed or sad at other behaviors) and so I do not let you hit
him. Both your feelings and my feelings are important, and each of us
can freely have his own." If the child were thus able to retain his own
organismic evaluation of each experience, then his life would 'become
a balancing of these satisfactions. Schematically he might feel, "I enjoy
hitting baby brother. It feels good. I do not enjoy mother's distress. That
feels dissatisfying to me. I enjoy pleasing her/' Thus his behavior would
sometimes involve the satisfaction of hitting his brother, sometimes the
satisfaction of pleasing mother. But he would never have to disown the

226 CARL R. ROGERS

feelings of satisfaction or dissatisfaction which he experienced in this
differential way.

F. The Development of Incongruence between Self and Experience

1. Because of the need for self-regard, the individual perceives his ex-
perience selectively, in terms of the conditions of worth which have come to
exist in him.

a. Experiences which are in accord with his conditions of worth are per-
ceived and symbolized accurately in awareness.

b. Experiences which run contrary to the conditions of worth are per-
ceived selectively and distortedly as if in accord with the conditions of
worth, or are in part or whole, denied to awareness.

2. Consequently some experiences now occur in the organism which are
not recognized as self-experiences, are not accurately symbolized, and are not
organized into the self-structure in accurately symbolized form.

3. Thus from the time of the first selective perception in terms of condi-
tions of worth, the states of Incongruence between self and experience, of
psychological maladjustment and of vulnerability, exist to some degree.

Comment. It is thus because of the distorted perceptions arising from
the conditions of worth that the individual departs from the integration
which characterizes his infant state. From this point on his concept
of self includes distorted perceptions which do not accurately represent
his experience, and his experience includes elements which are not in-
cluded in the picture he has of himself. Thus he can no longer live as a
unified whole person, but various part functions now become char-
acteristic. Certain experiences tend to threaten the self. To maintain
the self-structure defensive reactions are necessary. Behavior is regulated
at times by the self and at times by those aspects of the organism's
experience which are not included in the self. The personality is hence-
forth divided, with the tensions and inadequate functioning which ac-
company such lack of unity.

This, as we see it, is the basic estrangement in man. He has not been
true to himself, to his own natural organismic valuing of experience, but
for the sake of preserving the positive regard of others has now come
to f alsify^jpaie^pf the values he experiences and to perceive them only
in terms based upon ITfeff^value "to others. Yet this has not been a con-
scious choice, but a natural — and tragic — development in infancy. The
path of development toward psychological maturity, the path of therapy,
is the undoing of this estrangement in man's functioning, the dissolving
°f conditions of worth, the achievement of a self which is congruent

Therapy, Personality,, and Interpersonal Relationships 227

with experience, and the restoration of a unified organismic , valuing
process as the regulator of behavior. " ~~ " ~~~*"~ ^

G. The Development of Discrepancies in Behavior

1. As a consequence of the incongruence between self and experience
described in F9 a similar incongruence arises in the behavior of the in-
dividual.

a. Some behaviors are consistent with the self-concept and maintain and
actualize and enhance it.

(1) Such behaviors are accurately symbolized in awareness.

b. Some behaviors maintain, enhance, and actualize those aspects of the
experience of the organism which are not assimilated into the self-
structure.

(1) These behaviors are either unrecognized as self-experiences or
rceive^ z*n distorted or selective fashion in such a way as to be
congruent with the self.

H. The Experience of Threat and the Process of Defense

1. As the organism continues to experience, an experience which is in-
congruent with the self-structure (and its incorporated conditions of worth]
is subceived as threatening.

2. The essential nature of the threat is that if the experience were
accurately symbolized in awareness, the self-concept would no longer be a
consistent gestalt, the conditions of worth would be violated, and the need
for self-regard would be frustrated. A state of anxiety would exist.

3. The process of defense is the reaction which prevents these events
from occurring.

a. This process consists of the selective perception or distortion of the

^ experience and/or the denial to awareness of the experience or some

portion thereof, thus keeping the total perception of the experience

consistent with the individual's self -structure, and consistent with his

conditions of worth.

{J 4. The general consequences of the process of defense, aside from its
preservation of the above consistencies, are a rigidity of perception, due to
the necessity of distorting perceptions, an inaccurate perception of reality,
due to distortion and omission of data, and intensionality.

Comment. Section G describes the psychological basis for what are
usually thought of as neuroti^^beliaviors, and Section H describes the
mechanisms of these behaviors. From our point of view it appears more
f undajnental to think of defensive behaviors ( described in these two sec-
tions) and disorganized behaviors (described below). Thus the de-

228 CARL R. ROGERS

fensive behaviors include not only the behaviors customarily regarded as
neurotic — rationalization, compensation, fantasy, projection, compul-
sions, phobias, and the like — but also some of the behaviors customarily
regarded as psychotic, notably paranoid behaviors and perhaps catatonic
^states. /The disorganized category includes many of the "irrational" and
"acute" psychotic behaviors, as will be explained below. This seems to be
a more fundamental classification than those usually employed, and
perhaps more fruitful in considering treatment. It also avoids any con-
cept of neurosis and psychosis as entities in themselves, which we believe
has been an unfortunate and misleading conception.

Let us consider for a moment the general range of the defensive
behaviors from the simplest variety, common to all of us, to the more
extreme and crippling varieties. Take first of all, rationalization. ("I
didn't really make that mistake. It was this way. . . . ") Such excuses
involve a perception of behavior distorted in such a way as to make it
congruent with our concept of self (as a person who doesn't make
mistakes). Fantasy is another example. ("I am a beautiful princess, and
all the men adore me.") Because the actual experience is threatening to
the concept of self (as an adequate person, in this example), this ex-
perience is denied, and a new symbolic world is created which enhances
the self, but completely avoids any recognition of the actual experience.
Where the incongruent experience is a strong need, the organism
actualizes itself by finding a way of expressing this need, but it is per-
ceived in a way which is consistent with the self. Thus an individual
whose self-concept involves no "bad" sexual thoughts may feel or ex-
press the thought "I am pure, but you are trying to make me think
filthy thoughts." This would be thought of as projection or as a paranoid
idea. It involves the expression of the organism's need for sexual satis-
factions, but it is expressed in such a fashion that this need may be
denied to awareness and the behavior perceived as consistent with the
self. Such examples could be continued, but perhaps the point is clear
that the incongruence between self and experience is handled by the
distorted perception of experience or behavior, or by the denial of ex-
perience in awareness (behavior is rarely denied, though this is possible),
or by some combination of distortion and denial.

7. The Process of Breakdown and Disorganization

Up to this point the theory of personality which has been formulated
applies to every individual in a lesser or greater degree. In this and the
following section certain processes are described which occur _pnly when
certain specified, conditions are present.

l~r If the individual has a large or significant degree of mcongruencel
between self and experience and if a significant experience demonstrating [

Therapy., Personality., and Interpersonal Relationships 229

^
this incongruence occurs suddenly, or with a high degree of obviousness,

then the organism's process of defense is unable to operate successfully.

2. As a result anxiety is experienced as the incongruence is subceived.
The degree of anxiety is dependent upon the extent of the self -structure
which is threatened.

3. The process of defense being unsuccessful, the experience is accurately
symbolized in awareness, and the gestalt of the self-structure is broken by
this experience of the incongruence in awareness. A state of disorganization
results.

4. In such a state of disorganization the organism behaves at times in
ways which are openly consistent with experiences which have hitherto
been distorted or denied to awareness. At other times the self may tem-
porarily regain regnancy, and the organism may behave in ways consistent
with it. Thus in such a state of disorganization, the tension between the
concept of self (with its included distorted perceptions) and the experiences
which are not accurately symbolized or included in the concept of self,
is expressed in a confused^^grwicy, first one and jhen^the other supplying
the "feedback" by which the organism regulates behavior.

Comment. This section, as will be evident from its less exact for-
mulation, is new, tentative, and needs much more consideration. Its
meaning can be illuminated by various examples.

Statements 1 and 2 above may be illustrated by anxiety-producing
experiences in therapy, or by acute psychotic breakdowns. In the free-
dom of therapy, as the individual expresses more and more of himself,
he finds himself on the verge of voicing a feeling which is obviously
and undeniably true, but whicM is flatly contradictory to the conception
of himself which he has held.^fSee 62, pp. 78-80, for a striking verbatim
example of this experience.] Anxiety results, and if the situation is ap-
propriate (as described under /) this anxiety is moderate, and the result
is constructive. But if, through overzealous and effective interpretation
by the therapist, or through some other means, the individual is brought
face to face with more of his denied experiences than he can handle,
disorganization ensues and a psychotic break occurs, as described in
statement 3. We have known this to happen when an individual has
sought c 'therapy" from several different sources simultaneously. It has
also been illustrated by some of the early experience with sodium
pentathol therapy. Under the drug the individual revealed many of the
experiences which hitherto he had denied to himself, and which ac-
counted for the incomprehensible elements in his behavior. Unwisely
faced with the material in his normal state he could not deny its
authenticity, his defensive processes could not deny or distort the ex-
perience, and hence the self-structure was broken, and a psychotic break
occurred.

230 CARL R. ROGERS

Acute psychotic behaviors appear often to be describable as behaviors
which are consistent with the denied aspects of experience rather than
consistent with the self. Thus the person who has kept sexual impulses
rigidly under control, denying them as an aspect of self, may now make
open sexual overtures to those with whom he is in contact. Many of the
so-called irrational behaviors of psychosis are of this order.

Once the acute psychotic behaviors have been exhibited, a process
of defense again sets in to protect the organism against the exceedingly
painful awareness of incongruence. Here I would voice my opinion
very tentatively as to this process of defense. In some instances perhaps
the denied experiences are now regnant, and the organism defends itself
against the awareness of the self. In other instances the self is again
regnant, and behavior is consistent with it, but the self has been greatly
altered. It is now a self concept which includes the important theme, "I
am a crazy, inadequate, unreliable person who contains impulses and
forces beyond my control." Thus it is a self in which little or no con-
fidence is felt.

It is hoped that this portion of the theory may be further elaborated
and refined and made more testable in the future.

/. The Process of Reintegration

In the situations described under sections G and H, (and probably in
situations of breakdown as described under I, though there is less evidence
on this) a process of reintegration is possible, a process which moves in the
direction of increasing the congruence between self and experience. This
may be described as follows :

1. In order for the process of defense to be reversed — for a customarily
threatening experience to be accurately symbolized in awareness and as-
similated into the self -structure., certain conditions must exist.

a. There must be a decrease in the conditions of worth.

b. There must be an increase in unconditional self-regard.

2. The communicated unconditional positive regard of a significant
other is one way of achieving these"corrditions.

a. In order for the unconditional positive regard to be communicated,
it must exist in a context of empathic understanding.

fe. When the individual perceives such unconditional positive regard,
existing conditions of worth are weakened or dissolved.

c. Another consequence is the increase in his own unconditional positive
self-regard.

d. Conditions 2a and 2b above thus being met, threat is reduced, the
process of defense is reversed, and experiences customarily threatening
are accurately symbolized and integrated into the self concept.

Therapy, Personality, and Interpersonal Relationships 231

3. The consequences of 1 and 2 above are that the individual is less likely
to encounter threatening experiences; the process of defense is less frequent
and its consequences reduced; self and experience are more congruent; self-
regard is increased; positive regard for others is increased; psychological ad-
justment is increased; the organismic valuing process becomes increasingly
the basis of regulating behavior; the individual becomes nearly fully
functioning.

Comment. This section is simply the theory of therapy which we
presented earlier, now stated in a slightly more general form. It is in-
tended to emphasize the fact that the reintegration or restoration of
personality occurs always and only (at least so we are hypothesizing) in
the presence of certain definable conditions. These are essentially the
same whether we are speaking of formal psychotherapy continued over
a considerable period, in which rather drastic personality changes may
occur, or whether we are speaking of the minor constructive changes
which may be brought about by contact with an understanding friend
or family member.

One other brief comment may be made about item 2a, above. Em-
pathic understanding is always necessary if unconditional positive regard
is to be fully communicated. If I know little or nothing of you, and ex-
perience an unconditional positive regard for you, this means little be-
cause further knowledge of you may reveal aspects which I cannot so
regard. But if I know you thoroughly, knowing and empathically under-
standing a wide variety of your feelings and behaviors, and still ex-
perience an unconditional positive regard, this is very meaningful. It
comes close to being fully known and fully accepted.

Specification of Functional Relationships in the Theory of Personality

In a fully developed theory it would be possible to specify, with
mathematical accuracy, the functional relationships between the several
variables. It is a measure of the immaturity of personality theory that only
the most general description can be given of these functional relation-
ships. We are not yet in a position to write any equations. Some of the
relationships implied in section II may be specified as follows:

The more actualizing the experience, the more adient the behavior
(A5, 6).

The more numerous or extensive the conditions of worth, the greater
the proportion of experience which is potentially threatening (Fl, 2).

The more numerous or extensive the conditions of worth, the greater
the degree of vulnerability and psychological maladjustment (F3).

The greater the proportion of experience which is potentially threat-
ening, the greater the probability of behaviors which maintain and en-

232 CARL R. ROGERS

hance the organism without being recognized as self-experiences (Gla,
6).

The more congruence between self and experience, the more ac-
curate will be the symbolizations in awareness (Gla, and HI, 2, 3).

The more numerous or extensive the conditions of worth, the more
marked will be the rigidity and inaccuracies of perception, and the
greater the degree of intensionality (#4) .

The greater the degree of incongruence experienced in awareness,
the greater the likelihood and degree of disorganization (73) .

The greater the degree of experienced unconditional positive regard
from another, based upon empathic understanding, the more marked
will be the dissolution of conditions of worth, and the greater the pro-
portion of incongruence which will be eliminated (/2, 3 ) .

In other respects the relationships in section / have already been
specified in the theory of therapy.

Evidence. The first sections of this theory are largely made up of
logical constructs, and propositions which are only partly open to em-
pirical proof or disproof.

Section F receives some confirmation from Cartwright [9], and Diller
[14], Section H from Chodorkoff [10] and Cartwright [9], whereas
Goldiamond [22] introduces evidence which might modify the definition
of subception. Section / is supported by the evidence previously given
for the theory of therapy in Part I.

Because it is a closely reasoned and significant experimental testing
of certain of the hypotheses and functional relationships specified in this
portion of the theory, ChodorkofFs study [10] will be described briefly.
His definitions were taken directly from the theory. Defensiveness, for
example, is defined as the process by which accurate symbolizations of
threatening experiences are prevented from reaching awareness.

He concentrated on three hypotheses which may be stated in theoreti-
cal terms as follows :

1. The greater the congruence between self and experience, the less
will be the degree of perceptual defensiveness exhibited.

2. The greater the congruence between self and experience, the more
adequate will be the personality adjustment of the individual, as this
phrase is commonly understood.

3. The more adequate the personality adjustment of the individual
(as commonly understood), the less will be the degree of perceptual de-
fensiveness exhibited.

Thus it will be seen that he was testing one of the definitions of the
theory (Congruence equals psychological adjustment) against clinical
and common-sense reality. He was also testing one of the relationships
specified by the theory (Degree of congruence is inversely related to de-

Therapy, Personality, and Interpersonal Relationships 233

gree of defensiveness) . For good measure he also completes the triangle
by testing the proposition that adjustment as commonly understood is in-
versely related to degree of defensiveness.

He gave the following operational meanings to the essential terms :

1. Self is defined as a Q sort of self-referent items sorted by the in-
dividual to represent himself as of now.

2. Experience. An exact matching of the theoretical meaning with
given operations is of course difficult. Chodorkoff avoids the term "ex-
perience," but operationally defines it by an "objective description"
which is a Q sort by a clinician of the same self-referent items, this sort-
ing being based on a thorough clinical knowledge of the individual,
gained through several projective tests. Thus the total experiencing of
the individual, as distinct from the self-concept he possesses in aware-
ness, is given a crude operational definition by this means.

3. Perceptual defensiveness is defined as the difference in recogni-
tion time between a group of neutral words tachistoscopically presented
to the individual, and a group of personally threatening words similarly
presented. (The selection of the words and the technique of presentation
were very carefully worked out, but details would be too lengthy here. )

4. Personal adjustment as commonly understood was defined as
a combined rating of the individual by four competent judges, the rating
being based on biographical material, projective tests, and other infor-
mation.

These definitions provide an operational basis for four measures
entirely independent of one another.

Chodorkoff translates his hypotheses into operational predictions
as follows:

1. The higher the correlation between the individual's self-sort and
the clinician's sorting for his total personality, the less will be the differ-
ence in his recognition threshold between neutral and threatening words.

2. The higher the correlation between the self -sort and the clinician's
sorting for the total personality the higher will be the rating of personal
adjustment by the four judges.

3. The higher the adjustment rating by the four judges, the lower
will be the difference in recognition threshold between neutral and
threatening words.

All three of these predictions were empirically upheld at levels of
statistical significance, thus confirming certain portions of the theory.

This study illustrates the way in which several of the theoretical con-
structs have been given a partial operational definition. It also shows
how propositions taken or deduced from the theory may be empirically
tested. It suggests, too, the complex and remote behavioral predictions
which may be made from the theory.

234 CARL R. ROGERS

III. A THEORY OF THE FULLY FUNCTIONING PERSON

Certain directional tendencies in the individual (ID and 11-42) and
certain needs (IIC, D) have been explicitly postulated in the theory
thus far presented. Since these tendencies operate more fully under cer-
tain defined conditions, there is already implicit in what has been given
a concept of the ultimate in the actualization of the human organism.
This ultimate hypothetical person would be synonymous with "the goal
of social evolution," "the end point of optimal psychotherapy," etc. We
have chosen to term this individual the fully functioning person.

Although it contains nothing not already stated earlier under I and
II, it seems worthwhile to spell out this theoretical concept in its own
right.

A. The individual has an inherent tendency toward actualizing his
organism.

B. The individual has the capacity and tendency to symbolize ex-
periences accurately in awareness.

1. A corollary statement is that he has the capacity and tendency to
keep his self-concept congruent with his experience.

C. The individual has a need for positive regard.

D. The individual has a need for positive self-regard.

E. Tendencies A and B are most fully realized when needs C and D are
met. More specifically, tendencies A and B tend to be most fully realized
when

1. The individual experiences unconditional positive regard from sig-
nificant others.

2. The pervasiveness of this unconditional positive regard is made evi-
dent through relationships marked by a complete and communicated
empathic understanding of the individual's frame of reference.

F. If the conditions under E are met to a maximum degree, the in-
dividual who experiences these conditions will be a fully functioning person.
The fully functioning person will have at least these characteristics :

1. He will be open to his experience.

a. The corollary statement is that he will exhibit no defensiveness.

2. Hence all experiences will be available to awareness.

3. All symbolizations will be as accurate as the experiential data will
permit.

4. His self-structure will be congruent with his experience.

5. His self-structure will be a fluid gestalt, changing flexibly in the
process of assimilation of new experience.

6. He will experience himself as the locus of evaluation.

a. The valuing process will be a continuing organismic one.

Therapy, Personality, and Interpersonal Relationships 235

7. He will have no conditions of worth.

a. The corollary statement is that he will experience unconditional self-
regard.

8. He will meet each situation with behavior which is a unique and
creative adaptation to the newness of that moment.

9. He will find his organismic valuing a trustworthy guide to the most
satisfying behaviors, because

a. All available experiential data will be available to awareness and used.

b. No datum of experience will be distorted in, or denied to, awareness.

c. The outcomes of behavior in experience will be available to awareness.

d. Hence any failure to achieve the maximum possible satisfaction, be-
cause of lack of data, will be corrected by this effective reality testing.

10. He will live with others in the maximum possible harmony, because
of the rewarding character of reciprocal positive regard (IIClc) .

Comment. It should be evident that the term "the fully functioning
person" is synonymous with optimal psychological adjustment, optimal
psychological maturity, complete congruence, complete openness to ex-
perience, complete extensionality, as these terms have been defined.

Since some of these terms sound somewhat static, as though such
a person "had arrived," it should be pointed out that all the character-
istics of such a person are process characteristics. The fully functioning
person would be a person-in-process, a person continually changing.
Thus his specific behaviors cannot in any way be described in advance.
The only statement which can be made is that the behaviors would be
adequately adaptive to each new situation, and that the person would be
continually in a process of further self-actualization. For a more com-
plete exposition of this whole line of thought the reader may wish to see
my paper on the fully functioning person [64] .

Specification of Functions. Our present state of thinking can be
given in one sentence. The more complete or more extensive the condi-
tions Ely E2, the more closely will the individual approach the asymp-
totic characteristics Fl through FIQ.

Evidence. The evidence regarding outcomes of therapy is in a gen-
eral way confirmatory of the direction taken in this theory, though by
its very nature it can never be completely tested, since it attempts to
define an asymptote.

IV. A THEORY OF INTERPERSONAL RELATIONSHIP

The most recent extension of our theoretical constructs has been the
attempt to formulate the order which appears to exist in all interper-
sonal relationships and interpersonal communication. This formulation

236 CARL R. ROGERS

springs, as will be evident, primarily from the theory of therapy, viewing
the therapeutic relationship as simply one instance of interpersonal rela-
tionship. For clarity of presentation the conditions, process, and outcome
of a deteriorating relationship and a deepening or improving relationship
will be set forth separately. Actually these are two points or spaces on a
continuum.

A. The Conditions of a Deteriorating Relationship

For communication to be reduced, and for a relationship to deteriorate,
the following conditions are necessary:

1. A person Y is willing to be in contact with person X and to receive
communication from him. (Note: Y's characteristics do not need to be
specified, beyond saying that he is an "average person,53 with some malad-
justment, some incongruence, some defensiveness. The theory is stated
largely in terms of person X. )

2. Person X desires (at least to a minimal degree) to communicate to
and be in contact with Y.

3. Marked incongruence exists in X among the three following elements:

a. His experience of the subject of communication with Y. (Which may
be the relationship itself, or any other subject.)

b. The symbolization of this experience in his awareness, in its relation
to his self-concept.

c. His conscious communicated expression (verbal and/or motor) of
this experience.

Comment. If the discrepancy in 3 is a vs. b, c, then X is psy-
chologically maladjusted in this respect, and the immediate consequences
of the condition tend to be personal. If the discrepancy is a, b, vs. c,
then the state tends to be labeled deceit, and the immediate con-
sequences tend to be social.

The extreme of this incongruence, and hence one end point of the
continuum, would be a complete or almost complete incongruence or
dissociation between the experience, its cognitive meaning (symboliza-
tion ) , and its expression.

B. The Process of a Deteriorating Relationship

When the preceding conditions exist and continue, a process is initiated
which tends to have these characteristics and directions :

1. The communications of X to Y is contradictory and/or ambiguous,
containing

a. Expressive behaviors which are consistent with X's awareness of the
experience to be communicated.

Therapy, Personality, and Interpersonal Relationships 237

b. Expressive behaviors which are consistent with those aspects of the
experience not accurately symbolized in X5s awareness. (See IIG
above. )

2. Y experiences these contradictions and ambiguities.

a. He tends to be aware only of Bla, that is X's conscious communica-
tion.3

b. Hence his experience of X's communication tends to be incongruent
with his awareness of same.

c. Hence his response tends also to be contradictory and/or ambiguous,
his responses having the same qualities described for X in Bla, b.

3. Since X is vulnerable, he tends to perceive Y's responses as potentially
threatening.

a. Hence he tends to perceive them in distorted fashion, in ways which
are congruent with his own self -structure.

b. Hence he is inaccurate in his perception of Y's internal frame of
reference, and does not experience a high degree of empathy.

c. Because Y is perceived as a potential threat, X cannot and does not
experience unconditional positive regard for Y. (Note: thus X pro-
vides the reverse of the conditions for therapy as described in 1-43,
4,5.)

4. Y experiences himself as receiving at most a selective positive regard.

5. Y experiences a lack of understanding or empathy.

6. The more Y experiences a selectiveness of positive regard and an
absence of empathy, the less free he is to express feelings, the less likely
he is to express self -referent feelings, the less likely he is to be extensional in
his perceptions, the less likely he is to express incongruencies between self and
experience, the less likely he is to reorganize his self -concept. (Note: in gen-
eral, the process of personality changes as described in IB is reversed.)

7. Since Y is expressing less of his feelings, X is even more unlikely to
perceive Y's internal frame of reference with accuracy, and both inaccuracy
of perception and distortion of perception make defensive reactions on X's
part more likely.

8. Another characteristic which may exist, particularly if X's com-
munication is primarily of negative feelings, is that those aspects of ex-
perience which are not accurately symbolized by X in his awareness tend,
by defensive distortion of perception, to be perceived in Y.

9. If this occurs, Y tends to be threatened to the degree that these relate
to his own incongruences, and to exhibit defensive behaviors.

3 This is a crucial point. If Y is sufficiently open to his experience that he is
aware of X's other communication — described in Bib — then b and c below do not
follow, and his own response to X is clear and congruent. If in addition to his
awareness of all of X's communication he experiences an unconditional positive
regard for X, then this would become an improving relationship, as described in
sections D, E, and F which follow.

238 CARL R. ROGERS

C. The Outcome of a Deteriorating Relationship

The continuance of this process results in

1. Increased defensiveness on the part of X and Y.

2. Communication which is increasingly superficial, expressive of less of
the total individual.

3. The perceptions of self and others, because of the increased defensive-
ness, are organized more tightly.

4. Hence incongruence of self and expression remains in status quo, or
is increased.

5. Psychological maladjustment is to some degree facilitated in both.

6. The relationship is experienced as poor.

Comment on A, B, C. It may clarify this technical and theoretical
description of a deteriorating relationship to illustrate it from some com-
monplace experience. Let us, for example, take the relationship of a
mother, X, toward her child, Y. There is, of course, mutual willingness
to be in psychological contact. The mother feels "You annoy me because
you interfere with my career,35 but she cannot be aware of this because
this experience is incongruent with her concept of herself as a good
mother. Her perception of this experience in herself is distorted, becoming
"I am annoyed at this instance of your behavior. I love you but I must
punish you." This is an acceptable symbolization of her experience, and
it is this which she consciously communicates to the child.

But Y receives not only this conscious communication. He also ex-
periences (but tends to be unaware of) the expressive behaviors in-
dicating a more general dislike of himself. His response may be of several
sorts, but its essential characteristic is that it will express the incongruence
which her divided communication has set up in him. One possibility is
that he will experience himself as bad and unloved, even when his
awareness of his behavior is that he is "good." Hence he will act and
feel guilty and bad, even when behaving in an approved manner. This
type of response is threatening to the mother, because his behaviors
expressing badness and unlovedness threaten to bring into awareness
her own rejecting feelings. Consequently she must further distort her
perception of his behavior, which now seems to her "sneaky" or "hang-
dog" as well as being occasionally annoying. The more this cycle
continues, the less acceptance Y feels, the less adequately he can express
his feelings, the more difficult it is for his mother to achieve any empathic
understanding, the more completely the two are estranged in the rela-
tionship, the more maladjusted each becomes. It is the exact steps in
such a relationship which we have endeavored to describe in the three
foregoing sections — the conditions which bring it about, the process

Therapy^ Personality, and Interpersonal Relationships 239

by which deterioration takes place, and the outcomes of such a de-
teriorated relationship.

D. The Conditions of an Improving Relationship

For communication to increase, and the relationship to improve., the
following conditions are necessary:

1. A person, Y7, is willing to be in contact with person X7, and to receive
communication from him.

2. Person X' desires to communicate to and be in contact with Y7.

3. A high degree of congruence exists in X' between the three following
elements :

a. His experience of the subject of communication with Y/.

b. The symbolization of this experience in awareness in its relation to his
self -concept.

c. His communicative expression of this experience.

E. The Process of an Improving Relationship

1. The communication of X' to Y' is characterized by congruence of
experience, awareness., and communication.

2. Y' experiences this congruence as clear communication. Hence his
response is more likely to express a congruence of his own experience and
awareness,

3. Since X' is congruent and not vulnerable in the area related to his
communication, he is able to perceive the response of Y' in an accurate
and extensional manner, with empathy for his internal frame of reference.

4. Feeling understood, Y7 experiences some satisfaction of his need for
positive regard.

5. X' experiences himself as having made a positive difference in the
experiential field of Y'.

a. Hence reciprocally, X7 tends to increase in feeling of positive regard
for Y'.

b. Since X' is not vulnerable in the area of the communication, the
positive regard he feels for Y7 tends to be an unconditional positive
regard.

6. Y' experiences himself in a relationship which, at least in the area of
communication, is characterized by congruence on the part of X7, an
empathic understanding by X7 of the internal frame of reference, and an
unconditional regard. (See L43, 4, 5.)

a. Hence all the characteristics of the process of therapy (IB} are
initiated, within the confines of the subject of communication.

b. Because Y' has less need of any of his defenses in this relationship, any
need for distortion of perception is decreased.

c. Hence he perceives the communications of X7 more accurately.

240 CARL R. ROGERS

7. Hence communication in both directions becomes increasingly con-
gruent, is increasingly accurately perceived^ and contains more reciprocal
positive regard.

F. Outcomes of an Improving Relationship

The continuance of this process results in the following:
1. All of the outcomes of therapy (IC1 through 15) may occur, subject
to the time limitation of the relationship between X' and Y', and also to
the mutually understood limitations of the area of the relationship (e.g., it
may be mutually understood that it is only a lawyer-client relationship, or
only a teacher-pupil relationship, thus tending to exclude many areas of
expression and hence to that degree limiting the extent of the outcomes) .
Thus, within these limitations, the relationship facilitates improved con-
gruence and psychological adjustment in both X' and Yy.

G. A Tentative Law of Interpersonal Relationships

Taking all of this section, we may attempt to compress it into one over-
all law governing interpersonal relationships, specifying the functional rela-
tionship between the constructs. Here is such an attempt.

Assuming a minimal mutual willingness to be in contact and to receive
communications, we may say that the greater the communicated congruence
of experience^ awareness, and behavior on the part of one individual, the
more the ensuing relationship will involve a tendency toward reciprocal
communication with the same qualities, mutually accurate understanding
of the communications, improved psychological adjustment and functioning
in both parties, and mutual satisfaction in the relationship.

Conversely, the greater the communicated incongruence of experience,
awareness, and behavior, the more the ensuing relationship will involve
further communication with the same quality, disintegration of accurate
understanding, lessened psychological adjustment in both parties, and
mutual dissatisfaction in the relationship.

Comment. This is still a theory in the making, rather than a finished
product. It does not grow out of consideration of research data and
grows only partly out of experience. Basically, it is deduced from the
theory of therapy and projects into a new area a series of hypotheses
which now require confirmation or disproof. The evidence gained in such
studies should not only modify or confirm the theory of interpersonal
relationships but should reflexively throw new light on the theory of ther-
apy as well.

Evidence. It is believed that there is evidence from experience and
some research evidence concerning this theory. It seems preferable, how-
ever, simply to present it as a deduced theory.

Therapy, Personality, and Interpersonal Relationships 241

V. THEORIES OF APPLICATION

To spell out in detail the various theories of application which have
been partially developed, would be too repetitious of what has gone
before. Hence only a descriptive suggestion will be given in each area
of the aspects of theory which would be applicable.

Family life. The theoretical implications would include these:

1. The greater the degree of unconditional positive regard which the
parent experiences toward the child :

a. The fewer the conditions of worth in the child.

b. The more the child will be able to live in terms of a continuing
organismic valuing process.

c. The higher the level of psychological adjustment of the child.

2. The parent experiences such unconditional positive regard only to the
extent that he experiences unconditional self-regard.

3. To the extent that he experiences unconditional self-regard, the
parent will be congruent in the relationship.

a. This implies genuineness or congruence in the expression of his own
feelings (positive or negative).

4. To the extent that conditions 1, 25 and 3 exist, the parent will
realistically and empathically understand the child's internal frame of
reference and experience an unconditional positive regard for him.

5. To the extent that conditions 1 through 4 exist, the theory of the
process and outcomes of therapy (IB, C) , and the theory of the process and
outcomes of an improving relationship (I'VE, F) , apply.

Comment. Stated thus briefly, the applications to family life may
easily be misunderstood. For a presentation of these and related ideas,
the reader is referred to [65].

Education and learning. To the extent that education is concerned
with learnings which significantly influence behavior and facilitate
change in personality, then the conditions of therapy (1-4) and the
conditions of an improving relationship (IVD) apply. This leads, among
other things, to more realistic, accurate, and differentiated perceptions
(IC1, 2) and to more responsible basing of behavior upon these per-
ceptions (IC3, 10, 15).

Comment. Since a reasonably full statement of the theory of
facilitating learning has already been set forth [62, chap. 9], no at-
tempt will be made to spell it out in detail here, even though a number
of the terms and constructs in this earlier presentation are not precisely
those which are used here.

242 CARL R. ROGERS

Evidence. Several studies of the application of this theory to the
educational process have been made. Gross [26], Schwebel and Asch
[74]., Asch [3], and Faw [17, 18], supply evidence which in general is
confirmatory.

Group leadership. Building upon the postulate regarding the nature
of the individual (ID) and extending this to apply to groups, it has been
hypothesized that to the extent that a perceived leader provides the
conditions of therapy ( L43, 43 5 ) or of an improving relationship ( TVD } ,
certain phenomena will occur in the group. Among these are the follow-
ing: the perceptual resources of the group will be more widely used,
more differentiated data will be provided by the group, thinking and
perceptions will become more extensional, self-responsible thinking and
action will increase, a greater degree of distributive leadership will de-
velop, and there will be more effective long-range problem solving. All
of these consequences flow logically from the theory thus far presented.
In two major expositions [24, 23], Gordon has set forth carefully the
theory of application in this field, and it will not be repeated here. The
reader is referred to these presentations for more detail.

Evidence. The studies by Roethlisberger and Dickson [57], Coch
and French [11], Radke and Klisurich [53], Gordon, and others supply
some confirmatory evidence of different aspects of the theory.

Group tension and conflict. In serious situations of group conflict,
the conditions of a deteriorating interpersonal relationship (IV A)
usually exist. Drawing both from the theory of therapy and the theory
of interpersonal relationships, certain hypotheses have been formulated
in regard to such situations. Since these introduce a somewhat new
point, they will be formulated in more detail.

For our present purpose we may assume as given a group situation
in which the conditions of a deteriorating relationship (IV A) already
exist, with defensive behaviors and expressions being mutually increased
between X and Y and Z, different members of the group, or between
different subgroups represented by X, Y, and Z.

A. Conditions of Reduction in Group Conflict

Group conflict and tension will be reduced if these conditions exist.

1. A person (whom we term a facilitator) is in contact with X, Y,
andZ.

2. The facilitator is congruent within himself in his separate contacts
with X, Y, and Z.

3. The facilitator experiences toward X, Y, and Z, separately:

a. An unconditional positive regard., at least in the area in which the
members of the group are communicating.

Therapy, Personality, and Interpersonal Relationships 243

b. An em pat hie understanding of the internal frame of reference of
X, Y, Z, at least In the area in which the members of the group are
communicating.

4. X, Y, and Z perceive, at least to a minimal degree^ conditions 3a and
36. (This is generally because 3b is communicated verbally.)

B. The Process of Reduction of Group Conflict

If the above conditions exist and continue, then :

1. The various elements of the process of therapy (IB) take place to
some degree, at least within the area involved in the group communication.

a. One of the important elements of this process is the increase in dif-
ferentiated perceptions and in extensionality.

b. Another important element is the reduction of threat (see IBS, 8a)
in the experience of X, Y, Z.

2. Consequently the communications of Y to X or Z to X., are less de-
fensive, and more nearly congruent with the experience of Y, and with the
experience of Z.

3. These communications are perceived with increasing accuracy and
extensionality by X.

a. Consequently X experiences more empathic understanding of Y
and Z.

4. Because he is experiencing less threat from Y and Z and more
empathy with their internal frame of reference:

a. X now symbolizes in awareness incongruencies which formerly existed
between experience and awareness.

b. Consequently his defensive distortions of his own experience are
reduced.

c. Hence his communication to Y and Z becomes a more extensional ex-
pression of his own total experience in regard to the area of com-
munication.

5. The conditions now exist for the process of an improving relationship,
and the phenomena described in TVE occur.

Comment. A more general statement of the views presented here
theoretically will be found in two previous papers [63, 61]. This theory
is a deduction from the theory of therapy, and the theory of interpersonal
relationships.

Evidence. Although clinical evidence tends to confirm the theory in
small face-to-face groups, and Axline [5] has given an account of such a
clinical situation, there is as yet, I believe, no research evidence bearing
on this aspect of the theory. Particularly crucial and important from a
social point of view will be investigations involving different sizes of
groups. Even if the theory is fully confirmed in small f ace-to-face groups,
will it hold true in larger groups where communication is not face-to-

244 CARL R. ROGERS

face? There is also a question involving groups composed of spokesmen,
or representatives, where the individual feels that he cannot speak out of
his own experience and feeling, but only in a way dictated by his con-
stituents, who are not present. It is quite clear that the theory, as
formulated here, would not directly apply to this last type of situation.

THE THEORETICAL SYSTEM IN A CONTEXT OF RESEARCH

Our presentation of the theoretical system is completed. It is to be
hoped that the presentation has made it clear that this is a developing
system, in which some of the older portions are being formulated with
considerable logical rigor, while newer portions are more informal, and
contain some logical and systematic gaps and flaws, and still others (not
presented) exist as highly personal and subjective hunches in the minds
of members of the client-centered group. It is also to be hoped that it is
evident that this is a system which is in a continual state of modification
and clarification. Comparison of the theory as given above with the
theory of therapy and personality given in Client-centered Therapy in
1951 [62, chaps. 4, 11] or with the paper presented to the APA in 1947
[60] will show that although the major directions have not markedly
changed, there have been many changes in the constructs employed, and
far-reaching changes in the organization of the theory. This ongoing
process of revision is expected to continue.

The major usefulness of the systematic theoretical thinking, aside
from the personal satisfaction it has given, has been the stimulation of
research. In this respect there seems little doubt that it has had con-
siderable success. By and large the order of events seems to have been
this — clinical therapeutic experience, formulation of theory, research
which tests the theory, new aspects of experience perceived because of
the research, modification of the theory in the light of the new experience
and the research, further empirical testing of the revised hypotheses.

It would take too much space to review or even list the studies which
have been made. This would also be an unnecessary duplication since
Seeman and Raskin [77] have written a thoughtful analysis and criticism
of 55 of the research studies in therapy and personality which have been
stimulated by this point of view and completed during the years 1942-
195 1.4 Suffice it to say that clusters of research investigations have been
made around each of the following subjects of inquiry :

1. The events and process of therapy. Analysis of recorded thera-
peutic interviews in terms of theoretical constructs has been a major
tool here.

4 Since writing the above D. S. Cartwright has published: Annotated bibliog-
raphy of research and theory construction in client-centered therapy, /. counsel.
PsychoL, 1957, 4, 82-100.

Therapy, Personality , and Interpersonal Relationships 245

2. The results or outcomes of therapy. Tests of personality and
measures of different aspects of behavior have been the major in-
strumentation.

3. Investigation of personality theory. Hypotheses regarding per-
ception of self, others, external reality, and perceived locus of evaluation
have been investigated with a wide range of instruments,

4. Application of theory in specific fields. Investigations particularly
in the facilitation of learning and in group leadership.

Since 1951, many more studies have been completed in the out-
comes of therapy, an important collection of these being gathered in
Psychotherapy and Personality Change [70]. In these studies the prob-
lem of a control group is much more adequately handled than heretofore,
giving the findings a solidity which is noteworthy. If the reader wishes
to obtain a first-hand grasp of the way in which refinements of instru-
mentation and general scientific sophistication have developed in this
field, he should compare the seven studies of therapeutic outcome
published in the Journal of Consulting Psychology in 1949 (the entire
July issue, pp. 149-220) with the thirteen studies published in Psy-
chotherapy and Personality Change ( 1954) .

In addition to the many studies of outcome there are an increasing
number which have as their primary purpose the investigation of em-
pirical predictions made from personality theory. The study of Ghodor-
koff [10], already cited, is an excellent example of this group. There are
also studies now in progress which draw their hypotheses from an inte-
gration of the theory of therapy with a theory of perception or a theory
of learning. Such studies will, it is hoped, link the findings in the field
of therapy to the findings in older and more established fields of
psychology.

The bases of stimulation of research. There are, in the writer's
opinion, several basic reasons why this theoretical system has been help-
ful in giving impetus to a wide variety of research investigations.

The first is the orienting attitude mentioned in the first section of
this document, that scientific study can begin anywhere, at any level
of crudity or refinement, that it is a direction, not a fixed degree of in-
strumentation. From this point of view, a recorded interview is a small
beginning in scientific endeavor, because it involves greater objectification
than the memory of an interview; a crude conceptualization of therapy
and crude instruments for measuring these concepts, are more scientific
than no such attempts. Thus individual research workers have felt that
they could begin to move in a scientific direction in the areas of greatest
interest to them. Out of this attitude has come a series of instruments of
increasing refinement for analyzing interview protocols, and significant
> beginnings have been made in measuring such seemingly intangible con-

246 CARL R. ROGERS

structs as the self-concept and the psychological climate of a therapeutic
relationship.

This leads me to what I believe to be the second major reason for
the degree of success the theory has had in encouraging research. The
constructs of the theory have, for the most part, been kept to those which
can be given operational definition. This has seemed to meet a very
pressing need for psychologists and others who have wished to advance
knowledge in the field of personality but who have been handicapped by
theoretical constructs which cannot be defined operationally. Take, for
example, the general phenomena encompassed in such terms as the self,
the ego, the person. If a construct is developed — as has been done —
which includes those inner events not in the awareness of the individual
as well as those in awareness, then there is no satisfactory way at the
present time to give such a construct an operational definition. But by
limiting the self-concept to events in awareness, the construct can be
given increasingly refined operational definition through the Q technique,
the analysis of interview protocols, etc., and thus a whole area of in-
vestigation is thrown open. In time the resulting studies may make it
possible to give operational definition to the cluster of events not in
awareness.

The use of operationally definable constructs has had one other
effect. It has made completely unnecessary the use of "success" and
"failure33— two terms which have no scientific usefulness — as criteria in
studies of therapy. Predictions can instead be made in terms of opera-
tionally definable constructs, and these predictions can be confirmed or
disconfirmed, quite separately from any value judgments as to whether
the change represents "success" or "failure." Thus one of the major
barriers to scientific advance in this area has been removed.

A third and final reason for whatever effectiveness the system has
had in mediating research is that the constructs have generality. Because
psychotherapy is such a microcosm of significant interpersonal relation-
ship, significant learning, and significant change in perception and in
personality, the constructs developed to order the field have a high
degree of pervasiveness. Such constructs as the self-concept, or the need
for positive regard, or the conditions of personality change, all have
application to a wide variety of human activities. Hence such constructs
may be used to study areas as widely variant as industrial or military
leadership, personality change in psychotic individuals, the psychological
climate of a family or a classroom, or the interrelation of psychological
and physiological change.

The problem of measurement and quantification. I do not feel com-
petent to discuss, at a sophisticated level of statistical knowledge, the
problems of measurement which have been met by our group. This is

Therapy, Personality, and Interpersonal Relationships 247

best left to others. I will only mention three examples of the continuing
trend toward ever more refined quantification of the data of psycho-
therapy and personality.

The researches which have taken their start from client-centered
theory have significantly advanced the field of analysis of verbal
protocols. Working with recorded interviews, increasingly exact methods
have been devised, so that reliability of categorization is high, and very
subtle constructs, such as, for example, an "emergent self-perception"
can be objectified and measured. The attempt has been made by
Grummon [27] to integrate some of the methods we have developed with
the more formal methods of language analysis.

Other research workers have taken the Q technique as developed
by Stephenson [81], and have exploited it in a variety of ways. It has
been used to give an operational definition to the self-concept, to pro-
vide objectifications of a diagnostician's perception of an individual im-
mediately comparable to that individual's self-perception, to measure
the quality of a relationship as perceived by the two participants, and to
test a variety of hypotheses growing from personality theory.

Butler [6] has developed a new method for discovering the order
which exists in such material as interview protocols. A number of people
working with him have begun to apply this method — termed Rank
Pattern Analysis — to problems of complex analysis which hitherto had
been baffling.

Thus in a number of different areas the researches stimulated by
client-centered theory have not only contributed to the empirical base
of the theory, but have contributed to the development of methodology
as well. In principle there seems no limit to the refinement of measure-
ment in the areas covered by the theory. The major obstacle to progress
has been the lack of sufficient inventiveness to develop tools of measure-
ment adequate for the tasks set by the theory.

Incompatible evidence. Some of the evidence related to the theory
has been cited in each section. It will have been noted that nearly all
of this evidence has been confirmatory and that which is not confirming
has tended to be confused. There is almost no research evidence which
appears flatly to contradict the predictions from the theory.

Two related exceptions are the study reported by Carr [8], and
a portion of the study made by Grummon and John [28, also 37] which
is discussed by Vargas [85]. Briefly, the facts seem to be that Carr
and John had pre- and posttherapy projective tests analyzed by psy-
chologists who were basically diagnosticians. They found little or no
change in the degree of adjustment, in the projective material. In a
series of 10 cases, the John ratings as discussed by Vargas had a
significant negative correlation with counselor ratings. Yet when these

248 CARL R. ROGERS

same materials are analyzed "blind33 by therapeutically oriented re-
searchers (for example, Dymond) positive change is found, and the
correlation with counselor ratings is significantly positive.

The explanation suggested by Vargas is that the diagnostician tends
to think of adjustment as stability, a more or less fixed "level of de-
fense" which is socially acceptable. The therapeutically oriented worker
— especially if influenced by client-centered theory — tends to think of
psychological adjustment as an openness to experience, a more fluid
expressiveness and adaptiveness. Hence what the diagnostician perceives
as loss of control or even disorganization may be perceived by the
therapeutically oriented person as progress toward reduced defensiveness
and greater openness to experience. How deep this contradiction goes,
and its full implications, can only be evaluated in the light of further
research.

The main source of incompatible evidence is not research evidence,
but a clinical point of view. By and large the psychoanalytically oriented
Freudian group has developed, out of its rich clinical experience, a point
of view which is almost diametrically opposed to the hypotheses regard-
ing the capacities and tendencies of the human organism formulated
above in Dl, 2, 3, and also diametrically opposed to the theory of the
fully functioning person in III. Very briefly stated, the Freudian group,
on the basis of its experience, tends to see the individual as "innately
destructive53 (to use Karl Menninger's words) and hence in need of
control. To members of this group the hypothetical individual pictured
earlier under A Theory of the Fully Functioning Person is a psychopathic
personality, because they see nothing that would control him. The
hypothesis that self-control would be natural to the person who is with-
out defenses appears to them untenable.

In very much related fashion, the theory which Gordon and others
have formulated regarding group behavior and group leadership is
almost diametrically opposed to the Freudian theory in this respect.
Freud's statements that "groups have never thirsted after truth53 and
that "a group is an obedient herd which could never live without a
master3' suggests something of the deep discrepancy which exists between
the two views.

Though the psychoanalytic theory in these two respects is not sup-
ported by any research evidence, it nevertheless deserves serious con-
sideration because of the soil of clinical experience out of which it
originally grew. The discrepancy seems even more puzzling and challeng-
ing when it is realized that both the Freudian group and the client-
centered group have developed their theories out of the deep and
intimate personal relationships of psychotherapy.

It is my belief that the discrepancy can be understood in a way

Therapy, Personality, and Interpersonal Relationships 249

which leaves the client-centered theory intact, but this does not seem
to be the place for such a discussion. It seems best to present these
incompatible views for what they are, two theoretical stands which are
in flat contradiction on some basic points. Only new integrations of
theory and much deeper research investigations can resolve the difference.

A continuing program of theory and research. The theoretical
system and the research program which are connected with client-
centered therapy have grown from within themselves. This point can
hardly be overemphasized. The thought that we were making a start
on a theoretical system would for me have been a most distasteful notion
even as little as a dozen years ago. I was a practical clinician and held
(horrible dictu!} an open scorn of all psychological theory, as my early
students at Ohio State can testify. This was true even at the same time
that I was beginning to discern the orderliness which existed in the
therapeutic process. I like to think that the theoretical system and far-
reaching web of research which have developed, have grown in an
organic fashion. Each plodding step has simply been a desire to find out
this, a desire to find out that, a need for perceiving whatever con-
sistencies, or invariances, or order exists in the material thus far un-
earthed.

Consequently when I am asked, as I am in the outline suggested for
this paper, "the extent to which the systematic program has been
realized," I feel it is the wrong question for this system. I have no idea
what will be the ultimate realization of the living program which has
developed. I can see some of the likely next steps, or the current di-
rections, but have no assurance that these will be taken. We have con-
tinued to move in the directions which are experienced as rewarding, not
necessarily in those directions which logic points out. I believe this has
been the strength of the program, and I trust it will continue.

Thus I believe that we are likely to see progress in the following
directions, but I am not sure of any of them. It seems likely that further
moves will be made toward theory and research in the field of perception,
enriching that field by the insights gained in therapy, and being enriched
by the wealth of research data and theory in perception which can be
brought to bear in the refinement of the theories we are developing.
One such study now in progress, for example, is attempting to investigate
perceptual changes which occur during therapy. The measures range
from those entirely concerned with social perception — of people, of rela-
tionships— to those entirely concerned with the physical perception of
form, color, and line. Does therapy change only social perception, or
does it alter even the most basic perceptual processes? If not, where on
this continuum does change cease to occur?

I visualize the same type of rapprochement with learning theory,

250 CARL R. ROGERS

where in my judgment we have much to offer in the way of new di-
rections in that field, as well as being able to use much of the material
available there. It also seems likely that a number of the hypotheses we
are formulating may be tested in the laboratory, some on human and
some on animal subjects, thus linking the field of personality and therapy
with so-called experimental psychology. There seems no reason, for ex-
ample, why research on the establishment and consequences of condi-
tions of worth, as spelled out in this theory, might not be carried out on
higher animals, with a wider range of experimental conditions and more
adequate controls than could be achieved with human subjects.

I regard it as possible that there may be a closer linking of our
theory with the developing interest in creativity in the humanities and
social sciences generally, and I trust that this theory may provide a
number of relevant hypotheses for testing. I regard it as veiy likely that
the implications of this body of theory for industrial production will be
developed much more fully — the beginnings, as described by Richard
in Gordon's book [23], seem very exciting. I believe it is possible that
the near future may see a clear linking with the psychiatric group and a
testing of the theory in a wider variety of human disorders, with a re-
duction in the professional parochialism which has thus far kept the
medical group largely ignorant of the research in this field.

One direction which appears only theoretically possible is the ex-
ploitation in governmental affairs and international relations of some of
the implications of this theory. I do not regard this as likely in the near
future.

I suspect that the discovery and development of a contextual basis
for this theory in some form of existential philosophy will continue. The
general orientation of philosophical phenomenology is also likely to
continue to have its influence in this respect. These are some of the
potentialities for future development — rather grandiose, to be sure —
which I see. The extent to which any of them will organically grow is
a matter which demands a gift of prophecy I do not have.

Immediate strategy of development. To return, in closing, to the
much more immediate issues facing us in the systematic development of
the theory, I see several problems which have very high priority if our
general systematic thinking is to have a healthy development. I will list
these problems and tasks, but the order of listing has no significance,
since I cannot determine the priority.

L We are urgently in need of new and more ingenious tools of
measurement. Stephenson's Q technique [81] has been most helpful and
Osgood's method for quantifying semantic space [51] also seems promis-
ing. But most urgently needed of all is a method whereby we might
give operational definition to the construct experience in our theory, so

Therapy, Personality, and Interpersonal Relationships 251

that discrepancies between self-concept and experience, awareness and
experience, etc., might be measured. This would permit the testing of
some of the most crucial hypotheses of the theoretical system. To be sure,
some attempts have been made to approach such an operational defini-
tion, but the instrumentation is exceedingly cumbersome and admittedly
inadequate.

2. An increased amount of experience with individuals classed as
psychotic, and the testing of a variety of the theoretical hypotheses in
therapeutic work with this group and in research with psychotics as sub-
jects, would round out and enrich our systematic thinking in an area
in which it is at present inadequate. It would provide the type of ex-
treme reality test which is most helpful in the confirmation, modification,
or disproof of a theoretical system. There would seem to be no barriers
except practical ones to such a development.

3. An increased amount of experience and careful studies of hy-
potheses developed from the theory are needed in the area of group
relationships. Hypotheses regarding leadership, facilitation of learning,
and reduction of social conflict seem particularly fruitful to study. Here
again, the test of the theory at one of its deduced extremes would be most
helpful in confirming or revising its core.

4. Still another urgent need — no doubt quite evident to readers of
this presentation — is the translation of the present theory into terms
which meet the rigorous requirements of the logic of science. Although
progress in this direction has been made there is still a woefully long
distance to go. Such a development., carried through by competent
persons, would greatly sharpen the deductive hypotheses which might
be drawn from the system, and hence provide more crucial tests of it.

5. The final need I wish to mention may seem to some very con-
tradictory to the one just voiced. Personally I see it as a possible
evolutionary step, not as a contradictory one. I see a great need for
creative thinking and theorizing in regard to the methods of social
science. There is a rather widespread feeling in our group that the logical
positivism in which we were professionally reared is not necessarily the
final philosophical word in an area in which the phenomenon of sub-
jectivity plays such a vital and central part. Have we evolved the optimal
method for approximating the truth in this area? Is there some view,
possibly developing out of an existentialist orientation, which might
preserve the values of logical positivism and the scientific advances which
it has helped to foster and yet find more room for the existing sub-
jective person who is at the heart and base even of our system of science?
This is a highly speculative dream of an intangible goal, but I believe
that many of us have a readiness to respond to the person or persons who
can, evolve a tentative answer to the riddle.

252 CARL R. ROGERS

CONCLUSION

I find myself somewhat appalled at the length and scope of the ma-
terial which has been presented. I suspect the reader shares this feeling.
I can only say, somewhat apologetically, that I had not fully recognized
the ramifying pervasiveness of our theoretical thinking until I endeavored
to bring it all under one verbal roof. If many of the outlying structures
appear to the reader flimsy or unfit for occupancy, I hope that he will
find the central foundation, the theory of therapy, more solid. If to
some degree this formulation bestirs individuals to more activity in re-
search designed to prove or disprove these hypotheses, or to more
activity in building a better, more rigorous, more integrated theory, then
the group which is collectively responsible for the foregoing theories will
be fully satisfied,

REFERENCES

1. Anderson, R. An investigation of the relationship between verbal and
physiological behavior during client-centered therapy. Unpublished doctoral
dissertation., Univer. of Chicago, 1954.

2. Angyal, A. Foundations for a science of personality. New York:
Commonwealth Fund, 1941.

3. Asch, M. J. Nondirective teaching in psychology: an experimental
study. Psychol Monogr., 1951, 65, No. 4.

4. Assum, A. L., & Levy, S. J. Analysis of a non-directive case with
followup interview. /. abnorm. soc. Psychol., 1948, 43, 78-89.

5. Axline, Virginia M. Play therapy and race conflict in young children.
/. abnorm. soc. Psychol, 1948, 43, 300-310.

6. Butler, J. M. Rank pattern analysis of counseling protocols. Un-
published manuscript (mimeo.). Univer. Chicago Counseling Center, 1954.

7. Butler, J. M., & Haigh, G. V. Changes in the relation between self-
concepts and ideal concepts consequent upon client-centered counseling.
In [70, chap. 4].

8. Carr, A. C. Evaluation of nine psychotherapy cases by the Rorschach.
/. consult. Psychol, 1949, 13 (3), 196-205.

9. Cartwright, D. Self-consistency as a factor in affecting immediate
recall Unpublished manuscript (mimeo.). Univer. Chicago Counseling
Center, 1955.

10. Chodorkoff, B. Self-perception, perceptual defense, and adjustment.
/. abnorm. soc. Psychol, 1954, 49 (4), 508-512.

11. Coch, L., & French, J. R. P., Jr. Overcoming resistance to change,
Human Relat., 1948, 1, 512-532.

12. Gofer, C. N., & Chance, J. The discomfort-relief quotient in
published cases of counseling and psychotherapy. /. Psychol, 1950, 29,
219-224.

Therapy, Personality, and Interpersonal Relationships 253

13. Cowen, E. L., & Combs, A. W. Followup study of 32 cases treated
by nondirective psychotherapy. /. abnorm. soc. PsychoL, 1950, 45, 232-258.

14. Diller, L. Conscious and unconscious self -attitudes after success and
failure. /. Pers., 1954, 23, 1-12.

15. Dymond, Rosalind F. Adjustment changes over therapy from self-
sorts. In [70,, chap. 5].

16. Dymond, Rosalind F. Adjustment changes over therapy from
Thematic Apperception Test ratings. In [70, chap. 8].

17. Faw, V. A psychotherapeutic method of teaching psychology. Amer.
Psychologist, 1949, 4, 104-109.

18. Faw, V. Evaluation of student-centered teaching. Unpublished
manuscript, 1954.

19. Fiedler, F. E. A comparative investigation of early therapeutic
relationships created by experts and non-experts of the psychoanalytic, non-
directive and Adlerian schools. Unpublished doctoral dissertation, Univer.
of Chicago, 1949.

20. Fiedler, F. E. A comparison of therapeutic relationships in psycho-
analytic, non-directive and Adlerian therapy. /. consult. PsychoL, 1950, 14,
436-445.

21. Gallagher, J. J. The problem of escaping clients in non-directive
counseling. In W. U. Snyder (Ed.), Group report of a program of research
In psychotherapy. Psychotherapy Research Group, Pennsylvania State
Univer., 1953. Pp. 21-38.

22. Goldiamond, I. On the asynchrony between responses in a per-
ceptual experiment. Unpublished doctoral dissertation, Univer. of Chicago,
1954.

23. Gordon, T. Group-centered leadership. Boston: Houghton Mifflin,
1955.

24. Gordon, T. Group-centered leadership and administration. In C. R.
Rogers, Client- centered therapy. Boston: Houghton Mifflin, 1951. Chap. 8.

25. Gordon, T., & Cartwright, D. The effects of psychotherapy upon
certain attitudes toward others. In [70, chap. 1 1].

26. Gross, L. An experimental study of the validity of the non-di-
rective method of teaching. /. PsychoL, 1948, 26, 243-248.

27. Grummon, D. L. The use of language categories as a method for
study of personality in psychotherapy. Unpublished doctoral dissertation,
Univer. of Chicago, 1951.

28. Grummon, D. L., & John, Eve S. Changes over client- centered
therapy evaluated on psychoanalytically based Thematic Apperception Test
scales. In [70, chap. 11].

29. Haigh, G. V. Defensive behavior in client-centered therapy. /. con-
sult. PsychoL, 1949, 13 (3), 181-189.

30. Haimowitz, Natalie Reader, & Morris, L. Personality changes in
client-centered therapy. In W. Wolff (Ed.), Success in psychotherapy. New
York: Grime & Stratton, 1952. Chap. 3.

31. Haimowitz, Natalie Reader. An investigation into some personality
changes occurring in individuals undergoing client-centered therapy. Un-
published doctoral dissertation, Univer. of Chicago, 1948.

254 CARL R. ROGERS

32. Hanlon, T. E., Hofstaetter, P. R., & O'Connor, J. P. Congruence of
self and ideal self in relation to personality adjustment. /. consult. Psychol.,
1954, 18 (3) ,215-218.

33. Hartley, Margaret. Changes in the self-concept during psycho-
therapy. Unpublished doctoral dissertation, Univer. of Chicago, 1951.

34. Hilgard, E. R. Human motives and the concept of the self. Amer.
Psychologist, 1949, 4, 374-382.

35. Hoffman, A. E. A study of reported behavior changes in counseling.
/. consult. Psychol, 1949, 13, 190-195.

36. Hogan, R. The development of a measure of client defensiveness
in the counseling relationship. Unpublished doctoral dissertation, Univer. of
Chicago, 1948.

37. John, Eve S. Mental health and the principle of least effort. Un-
published doctoral dissertation, Univer. of Chicago, 1953.

38. Jonietz, Alice. A study of phenomenological changes in perception
after psychotherapy as exhibited in the content of Rorschach percepts. Un-
published doctoral dissertation, Univer. of Chicago, 1950.

39. Kauffman, P. E., & Raimy, V. C. Two methods of assessing thera-
peutic progress. /. abnorm. soc. Psychol., 1949, 44, 379-385.

40. Kelley, E. G. Education in communication. ETC, Summer, 1955,

12, 248-256.

41. Kessler, Carol. Semantics and non-directive counseling. Unpublished
master's thesis, Univer. of Chicago , 1947.

42. Kirtner, W. L. Success and failure in client-centered therapy as a
function of personality variables. Unpublished master's thesis, Univer. of
Chicago, 1955.

43. Lecky, P. Self -consistency: a theory of personality. New York:
Island Press, 1945.

44. Lipkin, S. Clients' feelings and attitudes in relation to the outcome
of client-centered therapy. Psychol. Monogr., 1954, 68, No. 1 (Whole No.
372).

45. Maslow, A. H. Motivation and personality. New York: Harper,
1954.

46. McCleary, R. A., & Lazarus, R. S. Autonomic discrimination with-
out awareness. /. Pers., 1949, 18, 171-1/9.

47. Mitchell, F. H. A test of certain semantic hypotheses by application
to client-centered counseling cases: intensionality-extensionality of clients in
therapy. Unpublished doctoral dissertation, Univer. of Chicago, 1951.

48. Mosak, H. Evaluation in psychotherapy: a study of some current
measures. Unpublished doctoral dissertation, Univer. of Chicago, 1950.

49. Muench, G. A. An evaluation of non-directive psychotherapy by
means of the Rorschach and other tests. Appl. Psychol. Monogr., 1947, No.

13, 1-163.

50. Nunnally, J. C. An investigation of some propositions of self-con-
ception: the case of Miss Sun. /. abnorm, soc. Psychol. , 1955, 50, 87-92.

51. Osgood, C. E. The nature and measurement of meaning. Psychol.
Bull, 1954, 49, 197-237.

Therapy, Personality, and Interpersonal Relationships 255

52. Quinn, R. D. Psychotherapists' expressions as an index to the quality
of early therapeutic relationships established by representatives of the non-
directive, Adlerian, and psychoanalytic schools. Unpublished doctoral dis-
sertation, Univer. of Chicago, 1950.

53. Radke, Marian, & Klisurich, Dayna. Experiments in changing food
habits. /. Amer. dietetic Ass., 23, 403-409.

54. Raimy, V. C. The self -concept as a factor in counseling and per-
sonality organization. Unpublished doctoral dissertation., Ohio State Univer.,
1943.

55. Raimy, V. C. Self reference in counseling interviews. /. consult.
PsychoL, 1948, 12, 153-163.

56. Raskin, N. J. An objective study of the locus of evaluation factor
in psychotherapy. Unpublished doctoral dissertation, Univer. of Chicago,
1949.

57. Roethlisberger, F. J., & Dickson, W. J. Management and the worker.
Cambridge, Mass.: Harvard Univer. Press, 1939.

58. Rogers, C. R. Clinical treatment of the problem child. Boston:
Houghton Miffin, 1939.

59. Rogers, C. R. Counseling and psychotherapy. Boston: Houghton
Mifflin, 1942.

60. Rogers, C. R. Some observations on the organization of personality.
Amer. Psychologist, 1947, 2, 358-368.

61. Rogers, C. R. Dealing with social tensions. New York: Hinds,
Hayden and Eldredge, 1948. 30 pp.

62. Rogers, C. R. Client-centered therapy. Boston: Houghton Mifflin,
1951.

63. Rogers, C. R. Communication: its blocking and its facilitation,
ETC, Winter, 1952, 9, 83-88.

64. Rogers, C. R. A concept of the fully functioning person. Un-
published manuscript (mimeo.). Univer. Chicago Counseling Center, 1953.
1953.

65. Rogers, C. R. The implications of client-centered therapy for family
life. Paper given to Chicago chapter of Int. Soc. Gen. Semantics, April,
1953.

66. Rogers, C. R. Persons or science: a philosophical question. Am.
Psychologist, 1955, 10, 267-278; also published in Cross Currents, Summer,
1953, 3, 289-306.

67. Rogers, C. R. The case of Mrs. Oak: a research analysis. In [703
chap. 15].

68. Rogers, C. R. Changes in the maturity of behavior as related to
therapy. In [70, chap. 13].

69. Rogers, C. R. This is me; the development of my professional think-
ing and my personal philosophy. Paper given at Brandeis Univer., Nov.,
1955.

70. Rogers, C. R., & Dymond, R. F. (Eds.) Psychotherapy and per-
sonality change. Chicago: Univer. Chicago Press, 1954.

256 CARL R. ROGERS

71. Rogers, Natalie. Changes in self-concept in the case of Mrs. Ett.
Personal Counselor, 1947, 2, 278-291.

72. Rogers, Natalie. Measuring psychological tension in non-directive
counseling. Personal Counselor, 1948, 3, 237-264.

73. RudikofT, Esselyn C. A comparative study of the changes in the con-
cept of the self, the ordinary person, and the ideal in eight cases. In [70,
chap. 11].

74. Schwebel, M., & Asch, M. J. Research possibilities in non-directive
teaching. /. educ. PsychoL, 1948, 39, 359-369.

75. Seeman, J. Counselor judgments of therapeutic process and out-
come. In [70, chap. 11].

76. Seeman, J. A study of the process of non-directive therapy. /.
consult. PsychoL, 1949, 13, 157-168.

77. Seeman, J., & Raskin, N. J. Research perspectives in client-centered
therapy. In O. H. Mowrer (Ed.), Psychotherapy: theory and research. New
York: Ronald, 1953.

78. Sheerer, Elizabeth T. The relationship between acceptance of self
and acceptance of others. /. consult. PsychoL, 1949, 13 (3), 169-175.

79. Snyder, W. U. An investigation of the nature of non-directive psy-
chotherapy. /. genet. PsychoL, 1945, 33, 193-223.

80. Standal, S. The need for positive regard: a contribution to client-
centered theory. Unpublished doctoral dissertation, Univer. of Chicago,
1954.

81. Stephenson, W. The study of behavior: Q-technique and its method-
ology. Chicago: Univer. of Chicago Press, 1953.

82. Stock, Dorothy. The self concept and feelings toward others. /.
consult. PsychoL, 1949, 13 (3), 176-180.

83. Strom, K. A re-study of William U. Snyder's "An investigation of
the nature of non-directive psychotherapy." Unpublished master's thesis,
Univer. of Chicago, 1948.

84. Thetford, W. N. An objective measure of frustration tolerance in
evaluating psychotherapy. In W. Wolff (Ed.), Success in psychotherapy,
New York: Grune & Stratton, 1952. Chap. 2.

85. Vargas, M. Changes in self -awareness during client-centered therapy.
In [70, chap. 10].

86. Zimmerman, J. Modification of the discomfort-relief quotient as a
measure of progress in counseling. Unpublished master's thesis, Univer. of
Chicago, 1950.

PERSONALITY THEORY GROWING FROM
MULTIVARIATE QUANTITATIVE RESEARCH

RAYMOND B. GATTELL

Laboratory of Personality
Assessment and Group Be-
havior, University of Illinois

Definition of the Approach {1} 257

Personality Research in Relation to the Two Basic Scientific Methods {1 +} . 261

The Logic of Factor Analytic Experiment {4-6} 265

The Conceptual Status and Interpretation of Factors {3-5+} 270

Classification of Factor Phenomena by Modality, Data, and Order {3 +, 2} . 273

The Present Status of Findings {8, 9} 278

Complex Function and Configural, Type Prediction from Source Traits {2+} 284
The Evidence for Motivational and Dynamic Factors; Ergs and Sentiments

{2 +,9} 290

The Dynamic Calculus of Ergic Strengths, the Self Sentiment, Conflict, and

Integration {2+} 296

Summary and Systematic Analysis of the Present Formulation {2-12} . . . 303

Appendix: The Concept of Variable Density and Factor Order 318

References 322

DEFINITION OF THE APPROACH

The maturity of theoretical developments may be tested by two
touchstones. First, a scientific system is generally more mature when its
concepts arise from specially developed operations and techniques other
than those available to everyday observation and to the layman. Sec-
ondly, theory is more mature if we can point to ensuing predictive and
controlling powers which are real enough to have led to potent tech-
nologies, recognizable in specially developed social institutions,

By these touchstones, "personality theory55 ranges more widely in de-
velopmental level than do most other areas of psychological theory, pre-
senting examples from rarefied heights and from degraded depths of
scientific acceptability and status. On the one hand, there is voluminous
theory — principally in the clinical area — based on no better methods of

257

258 RAYMOND B. GATTELL

observation than have been available for centuries, entangled in verbal
stereotypes that are almost certainly false or purely local in reference,
intuitive in observation, inexplicit as to assumptions, and in general, not
precisely, operationally based or confirmable. From this level of scientific
poverty it rises, on the other hand, to rational, objective, quantitative,
and intricately developed concepts which can truly be said to surpass, in
both complexity of testable theory and effectiveness of technological re-
sults, such neighboring fields as, say, learning theory and group dynamics.
The present essay is concerned exclusively with the kind of person-
ality theory which has developed out of quantitative and objective meth-
ods, whether that is based upon clinical, abnormal data, upon social or
educational fields of observation, or upon laboratory and physiological
study of the normal individual. It is also demarcated from neighboring
developments by emphasis on multivariate analytic experiment rather
than on manipulative univariate experiment. This distinction will be
drawn more clearly in a moment; at the outset, let it be said that in
intention researchers in the present area are aiming at an experimentally
based personality theory and that the emphasis which has developed on
multivariate rather than on the traditional, controlled, univariate ex-
perimental method is considered only an intelligent strategic adaptation
to the needs of personality investigation at its present stage. It is con-
tended that much effort has been relatively wasted in unimaginative
application of classical experimental design by psychologists of impec-
cable scientific aspirations, who have failed to perceive that in psychology
(as distinct from the physical sciences) we encounter a situation and
kind of data to which classical design is not the best approach. In par-
ticular, the new multivariate experimenter contends that classical uni-
variate experiment has insufficiently realized: (1) that in psychology
(compared with physics), special steps must be taken to isolate organi-
cally unitary and unique behavior structures, i.e., "significant variables,"
before univariate experiment can be strategically applied, ( 2 ) that where
so many variables exist (even if restricted to "significant" ones) the
multivariate approach is far more economical and powerful in mapping
those systematic relations among variables from perception of which
the better-adapted hypotheses and models will arise.

This contrast with univariate experiment is not the whole story
regarding the character of the multivariate experimental approach con-
sidered in this essay. Indeed, in later stages, it has lost some of the char-
acters which initially distinguished it from the univariate experimental
tradition; notably, it has begun to manipulate variables (thus introduc-
ing dependent-independent variable concepts), although in a framework
of simultaneous operation with many variables. But it has also gathered
new characters of its own through growing into fresh branches of highly

Personality Theory from Quantitative Research 259

technical specialized experiment and conceptualization which have no
counterparts in univariate experiment.

In conformity with the plan suggested for the present contributions
on systematic viewpoints we shall begin with background factors and
orienting attitudes. Fortunately, in the present case these are so well
known that only a very brief sketch need be given. The main historical
roots of the method lie directly in the study of individual differences,
though in the last decade it has emancipated itself completely from this
restriction and has been largely concerned with lifting the study of struc-
ture and process to new technical levels. In social science generally, the
multivariate method began along with statistical method, or at least,
with the second covariational-analysis phase of the development of
statistics by Galton and Karl Pearson. In psychology, it grew to a lusty
adolescence in the study of individual differences in ability and school
achievement. This growth began with Spearman's and Burt's attempts
fifty years ago to place intelligence testing on a firm basis of theory and
continued through Thurstone's development of multifactor analysis. In
some isolated backwaters of academic teaching, the multivariate ap-
proach is still seen in these terms of individual differences, of nonmanip-
ulative experiment, of the merely economical objective of finding con-
venient "dimensions," and of restriction largely to educational and
cognitive psychological problems.

Actually, multivariate methods, of which factor analysis remains the
chief development, now handle far more issues than this and have as
much to contribute in personality, learning, and motivation study as in
the field of abilities. They offer as much in general experimental design
as in the psychometric study of individual differences. It is a truism of
scientific history that classifications which appear early are rarely those
which are ultimately realized through the logical, inherent characters
of the methods concerned. The approach defined here may be seen
historically as beginning with the structural and taxonomic problems of
classifying abilities. Yet in terms of its inherent, logical nature and the
real applications indicated for the future, multivariate analysis must be
seen as one of the two main experimental methods available in science
generally. Incidentally it may be a matter of justifiable pride to psycholo-
gists that although multivariate analytical methods are being used,
crescendo, in physiology, medicine, meteorology, and sociology, they
were largely developed within psychology (as the univariate methods
were within the older physical sciences) .

Before proceeding to bring out more explicitly the procedures and
assumptions of the multivariate approach, we may help the reader to-
ward perspective by giving some indication of substantive content, and
also of the relation of this contribution to others in the same series.

260 RAYMOND B. CATTELL

The theories developed here are flanked, on the motivational side, by
Miller's development of conflict theory (vol. 2) and Rapaport's account
of psychoanalytic structural and dynamic concepts. The factor analytic
account of unitary drives needs to be aligned with Morgan's physiological
picture (vol. 1). Our discussion of factors of temperament and the
methodology of psychological genetics should be brought into relation
with studies of human heredity in various fields (e.g., Kallmann). Our
general structural theory of personality has affinities to and differences
from the stimulus-response formulations of Hull and Spence (vol. 2) and
the more clinical analysis by Murray. On the social side, our mathe-
matical models for attitudes and for roles can be related to the contribu-
tions of Katz and Stotland and of Newcomb, respectively.

As these relationships are studied, it will become evident that the
present approach is not so much concerned with a theory, i.e., with a
particular set of constructs and concepts in, for example, personality and
motivation, as with many possible theories, all dependent upon the
resolving power of a particular methodological approach. Its unity is not
that of adherence to conceptual beliefs but of the natural integration
which exists in findings from a particular method and model, flexibly
applied and checked against other methods where possible. Nevertheless,
we admit a certain attachment to the theories per se, and certainly we
concentrate on the theories to the extent that this essay is not concerned
with all derivatives of the multivariate method, e.g., in group dynamics,
culture-pattern psychology, physiology, but with those developed in
personality and motivation. When concentrating on the theories, how-
ever, it is important to distinguish them from superficially similar notions,
often with similar names (e.g., Freud's notion of ego strength or Me-
Dougall's concept of the self-regarding sentiment) , which do not arise
from this mathematical model or bear the hallmark of statistical pre-
cision in measurement which the present concepts always imply.

To orient the reader from this point on, the author should state
that despite his intention to follow the excellent editorial outline sug-
gested for all contributions, he has been unable entirely to adapt the
present systematic material to the rubies indicated. Thus, after the above
statement of background the outline proceeds to the structure of the
system, particularly the systematic independent, intervening, and de-
pendent variables. This sequence is ill-adapted to the present case be-
cause initially there are no dependent and independent variables. At
least in the factor analytic method as used by Spearman, Burt, and
Thurstone all the individual difference measurement variables stand on
an equal footing.

Accordingly, it has seemed best to follow an order of exposition which
most clearly develops an understanding of the dependence of ideas upon

Personality Theory from Quantitative Research 261

procedures (some of which will probably not be initially known to the
reader) and then, in a final section, explicitly to summarize our position
in terms of the issues raised by the outline.

PERSONALITY RESEARCH IN RELATION TO THE TWO BASIC
SCIENTIFIC METHODS

After attempting to demarcate the special character and intention of
this particular contribution to personality theory, we now examine more
closely its chief instrument — multivariate analytical experiment. Since
the terms univariate and multivariate may not be understood in the same
sense by all, a brief comparative analysis, of these and other allegedly
distinct methods used in personality research, is necessary before empirical
findings, and the psychological concepts ensuing, can be properly focused.

Actually, it is common to hear three methods mentioned in per-
sonality research: the clinical method, the controlled experimental
method, and the multivariate analytic method — besides special emphases
and approaches cutting across these, such as the anthropological, the
physiological, etc. In the controlled, manipulative classical or univariate
experimental method, the independent variable is manipulated, or
allowed to alter, while all other variables are considered to be controlled,
except for the changes in the single dependent variable, which are
recorded. (Hence univariate, for occasionally the independent variable is
multiplied to two or three, as in the Fisherian factorial design.) Ex-
cept in a purely positivistic theoretical framework, the empirical in-
dependent variable is understood by the "classical" experimenter to
represent a systematic independent variable — a concept or construct
which he has postulated to be so represented. But, for the moment, we
shall set aside what the experimenter thinks he is doing — for this can be
differently conceived — and ask only what distinguishes univariate and
multivariate experiment in terms of what the experimenter actually does.

Before proceeding, we must deny the third approach, the "clinical
method," any status as a fundamentally distinct method. The only
logically possible treatments of relations among variables are in pairs
and sequentially, as in univariate experiment, or in large numbers and,
usually, without knowledge of sequence, as in multivariate experiment.
The clinician is generally a multivariate experimenter, who abstracts laws
and concepts from observing ("globally" or by "gestalts" as he might
say) simultaneous changes in a large number of uncontrolled variables.
Fundamentally he does exactly — or perhaps we should say inexactly —
what the factor analyst or multivariate experimenter does, but he does it
without the benefit of precise instrumental measurement or explicit
correlational procedures (or other mathematical treatment of functional

262 RAYMOND B. CATTELL

relations). His intuitions about functional unities are thus approxima-
tions to the analyst's independent factors, and his statements about
mutual influences of factors are made without benefit of an F or t test.
The "clinical method" does exist and function usefully as a rough
"reconnaissance" form of the basic multivariate experimental method.
Claims that it is anything other than this confuse the clinical method
as a subdivision of therapy, which it undoubtedly is, with an independent
scientific method, which it undoubtedly is not. For such claims give
merely local skills and methodological accretions grown up around
clinical practice the status properly due only to a fundamental difference
of design.

One cannot avoid the judgment that the valuable contribution of
clinical practice as an exploratory method has lately been more than
offset by its tendency to choke the growth of sound, checkable personality
theory in a rank weedy jungle of facile verbal concepts. Where quantita-
tive and computational checks are not possible — or, at least, are avoided
by the formulation of "theories" — the theoretical field becomes a mere
playground for persons of high fluency. If we apply the test of maturity
of theory suggested above — the production of an effective technology —
then the few existing examinations, notably by Kelley and Fiske [67],
Meehl [76], and Eysenck [50], showing that clinical psychology is in-
distinguishably above chance in diagnosis or therapy, leave us no con-
clusion but that purely clinically derived theory is in a bad way.

Ironically, however, when clinicians or others have tried to put their
house in order and to extract the true metal of science from the ores
in which clinical data are richer than laboratory data, they have reverted
to classical instead of that multivariate experiment which is intrinsic to
"clinical method" and the potential source of its greatest contribution.
This failure is rooted partly in education — the rarity of coordination of
mathematical and clinical training — and perhaps partly in temperament.
Lack of foresighted handling of the clinical research training programs
in this respect is likely to be responsible for our knowing in 1970 laws
which we might have known and applied in 1960. For, before controlled
experiment can go to work on the relation of, say, superego strength to
early family attitudes, or the changes of "free" and "bound" anxieties
under treatment by ataractic drugs, multivariate research must first
substantiate the existence of a unitary factor of superego strength, show
tests which measure it with defined concept validity and reliability, and
discover whether anxiety in fact falls into one, or two, or more, in-
dependent sets of manifestations.

The emphasis in the present contribution on theory derived from
multivariate rather than univariate quantitative research is, in summary,

Personality Theory from Quantitative Research 263

justified both by the historical situation and by the intrinsic logic of
method, as follows:

First, by the purely sociohistorical fact mentioned above, that the
method has been subject to gross and untimely neglect in relation to
realistically evaluated potential contribution. It has been neglected be-
cause, at least in Germany and America, those interested in objective
scientific research in psychology have largely been conservatively trained
in the half-truth that in psychology, as in physics, science consists of
controlled experiment. ' The clinicians who had the courage to break
away from this tradition realized that the more important emotional
situations could not be used in controlled experiments with man. A rigid
adherence to laboratory experiment would lead to the restriction of data
to such specialized but "trivial" fields as perception, the psychophysiology
of reflexes, or the sense organs, or to experiments on the emotions of
animals, which could never be applied, except by uncertain analogy, to
the personalities of human beings.

Secondly, at this primitive stage of personality research especially,
the multivariate method offers a swifter and surer approach to the
significant variables for controlled experimentation. In personality, as in
psychology and the life sciences as a whole, the investigator has an in-
finite array of variables from which to choose. It is not surprising — and
is perhaps a comment on our ways of striving for originality — that one
and the same empirical (not conceptual) variable rarely gets confirm-
atory investigation by as many as two psychologists. Apparently, there
are at least as many variables claimed to be of outstanding significance as
there are psychologists.

One of the common schemata underlying presentations in this book
is the statement of independent, intervening, and dependent variables in
each field. It has been editorially suggested that an independent (or
dependent) variable should be further considered in experimental,
mathematical, and ideational (systematic) senses. This initial clarification
is best adapted to the univariate methodology from which it was de-
rived; in the multivariate field it needs further structuring. A factor is
both a systematic, conceptual independent (or dependent) variable and
an intermediate variable. The strict multivariate methodologist is un-
likely, indeed, to introduce any conceptual intermediate variable that
is not first revealed as a unitary factor. However, the proof that a uni-
tary entity exists, and that it is therefore profitable to begin setting up
hypotheses about it, as a unitary concept, may occur years ahead of the
confirmation of what the entity is.

The contention of the multivariate analyst is that too many psy-
chologists have immaturely "jumped the gun" by imitating the univari-

264 RAYMOND B. CATTELL

ate experiment of physics in psychology, without regard to the different
stages and natures of these sciences. Greater shrewdness might have
foretold the impotence which these incontinent procedures have demon-
strated during more than half a century. It would have— and now em-
phatically has — indicated that a better strategy is to reduce the chaos
of infinite possible variables to more tractable and significant numbers
and natures by factor analysis before much hypothesis formation and
manipulative experiment begins. A decade devoted largely to systematic,
cooperative studies of this kind would not now be out of place in per-
sonality study, or indeed in learning, physiological, and social psychology.
The third and last relation to be emphasized between multivariate
and univariate methods justifies greater resort to the former not only
because of aptness to the present developmental phase, but generally. This
is the argument from research economy and certainty of inference. It
springs from three sources :

1. One multivariate research with, say, 30 variables yields evidence
on (30X29)/2 = 435 relationships, with only fifteen times the ex-
perimental work required in one univariate experiment. Consequently it
achieves the results of 435 univariate experiments with about one-
thirtieth of the expenditure.

2. The relationships are determined under conditions in which all
variables are allowed to vary over their full range together. Consequently
one does not have the uncertainty, which occurs in trying to make in-
ferences from many univariate experiments, as to possible interaction
effects lost through the controlled situation or as to corrections necessary
in integration because the different univariate relations have been found
on diverse samples.

3. The hypothesis being tested is made more determinate through
being represented by a factor measurement based on several empirical
independent variables instead of one only. For example, an investigator
may set out to test the hypothesis that rigidity is related to rate of con-
ditioning and state that operation X defines operationally his hypothesis
or concept of the nature of rigidity, i.e., X is the empirical independent
variable defining his systematic independent variable, rigidity. But fac-
tor ^ analysis might show, as it frequently has, that only one-third of the
variance of X is accounted for by a rigidity factor and that the rest is
equally determined by two other factors, say, intelligence and fatigue.
It may take several blind conceptual-trial-and-error studies before the
univariate experimenter hits on a better variable to represent rigidity
(as judged by more consistent or positive results), and during that time
his^ conclusions could just as well be statements about relations of con-
ditioning to intelligence or fatigue as to rigidity. A factorist would first
determine the factor structure and then tie the factor down by several

Personality Theory from Quantitative Research 265

operational representatives. For it is rare to find a complex concept that
can be represented by a single operation, and it is still more rare for a
univariate experimenter to land on it at the first attempt.

Against these substantial advantages two shortcomings can be
charged to the multivariate method, ( 1 ) that it only handles linear re-
lationships, ( 2 ) that it omits time sequence and therefore does not per-
mit unambiguous causal inference. (More narrowly stated, it deals with
response-response relations rather than stimulus-response.) The first is
true, but it is generally desirable to observe any relation in the approxi-
mate, linear form before proceeding to more complex functions. How
many relations do we yet know of in psychology involving a law that is
indubitably different from one of simple proportion? And are not most
controlled experiments content with an analysis of variance significance
test, proving nothing at all about the form of the relation? As to the
second, it rests largely on lack of reading in multivariate methods. The
condition-response factor design [23, 42] systematically investigates the
relation of controlled changes of stimulus to response; P technique [35]
and incremental R technique use factor analysis over time intervals
rather than in instantaneous, nonsequential analysis.

These later, more developed multivariate designs permit causal in-
ference about interaction of factors to be drawn from the same experi-
ment as that which structures the variables into factors, as will be seen
in examples in the following sections. They retain, however, the ad-
vantage that manipulative control of most variables is not necessary, as
it typically has to be in most univariate experiment. Instead of "isolating
by control," the multivariate experiment allows nature to vary as it will
(often producing effects we should not dare to duplicate in human ex-
periment) and then isolates by superior statistical analysis what cannot
be isolated by physical manipulation. For example, one might be in-
terested in the effects of Group Morale Factor 2 [41] upon individual
responses expressed in murder rates. Fortunately we are spared responsi-
bility for the latter because we do not know how to manipulate factor 2,
but we can accurately measure its changes and investigate the relations
accordingly. The wider realm of multivariate experimental design can
be read about elsewhere [23, 28, 58, 90], so we shall now confine our-
selves to the relevant essentials of the factor analytic model.

THE LOGIC OF FACTOR ANALYTIC EXPERIMENT

We need not deal with the mathematics and the computational pro-
cedures of factor analysis here [see 23, 58, 90], but its logic should be
briefly stated. Any of the standard factor analytic procedures will reduce
the variance on a large member n of individual variables to variance on

266 RAYMOND B. CATTELL

a small number of common factors k plus variance on n specific factors.
Thereafter the score P of a person i on a specific variable ; can be es-
timated by the specification equation :

Pji = SjiFu + SjtFzi + * • • + SjhFki + SjFj*

where the /s are the situational indices or loadings, obtained by factor-
ing the correlation matrix for the n variables, and the F's are the
strengths of the endowments of the individual i in the various factors.
Factors 1 to n are factors common to this and other performances,
whereas factor ; is specific to this particular response. The factor matrix,
obtained by factor analytic procedures from the correlation matrix,
gives us all we need for the above general equation. Each row of the
matrix gives the set of /s for estimation of the given variable, and each
column, presenting a factor, shows which variables need to have their
weighted scores added together to give an estimate of that factor for any
individual.

It will be noted in passing that this formulation again transcends, or
requires a new view of, the reduction of scientific systems to independent
and dependent variables and intermediate variables or constructs. For
the initial variables are ( at least in timeless, instantaneous factor analysis )
both the independent variables from which the construct — the factor — is
inferred, and the dependent variables predicted from these intermediate
variables, in the specification equation.

The majority of factor analytic researches are not carried out with
the object of proceeding to actual specification equation computations
but rather with the general scientific aim of determining the number and
nature of the psychological factors at work in a given phenomenal area.
At this level, issues have been much confused by difference of purpose
between mathematical statisticians and psychological researchers. The
mathematical statisticians are content if they can find a reduced number
of orthogonal factors which will reproduce the correlations, and if pos-
sible also the given scores, within the given experiment. But psychologists
are concerned to know that they have found the correct number of
factors and that they have the correct nature (pattern) for each factor,
in terms of other experiments beyond the one in question, i.e., in terms of
general scientific concepts. Consequently psychologists do not see ad-
vantage in the mathematical neatness of orthogonality; they positively
reject it, because it is highly probable that all factors in the same universe
have interaction and are likely to be somewhat correlated among
themselves.

The mathematician knows many — indeed an infinite number — of
combinations of numbers and natures of factors that will reproduce the

Personality Theory from Quantitative Research 267

given variable correlations and scores, and if he prefers any one, it will
be for mathematical neatness. The psychologist wants conditions for de-
termining a unique solution, i.e., a fixed number of factors rotated to
one fixed position, and he is more concerned that this unique interpre-
tation fit the interpretation of other experimental matrices than that it
fit certain concepts of mere mathematical convenience within one matrix.
The pursuit of this latter aim is tied up technically with development of
(1) communality estimation theories, (2) the invention of formulas for
standard errors for factors and loadings, and (3) the determination of
unique rotation positions by simple structure, criterion rotation, and
parallel profiles [1,7, 23, 29, 31, 44, 81, 90, 95], In these, statistical logic
has had at times to limp along with the help of a crutch derived from
empirical generalizations; but as of 1958, the major problems have been
overcome just sufficiently, though not always to the satisfaction of the
theoretical statistician. That is to say, researchers will in general now
agree on how many factors there are, and the "payoff3 of arranging
findings from many studies side by side shows that simple structure is
capable of revealing the same factor patterns from different, independent
experimental studies. An essential part of this completion of adequate
techniques has been the development of factor-matching indices, such as
the recent formula by Gattell and Baggaley [31], which permit us to
give fiducial limits to the goodness of a given matching of factors from
one study to another. With improved techniques for obtaining unique
resolution into factors, and improved methods of checking factors from
study to study, it has been possible to demonstrate the in variance of 10
to 20 personality and ability factors.

The logic of the resolution of variance on many variables into a set
of unique common factors, specifics, and error, is the same for all factor
analytic designs, regardless of experimental setting. But the uses of factor
analysis in different contexts of stimulus, response, and organism, and the
scientific meaning of the factors derived therefrom, fall basically into six
distinct experimental designs — actually a set of three basic designs, each
analyzable in two different ways [22, 23]. The three basic designs arise
from the nature of behavioral measurement. Any behavioral measure-
ment is defined and tagged by five referents: a particular organism
making the response, a particular stimulus situation in which the re-
sponse occurs, a particular moment in time, a particular point in space,
and a particular observer [22]. Setting aside the two last as irrelevant to
the basic designs, we have three characteristics, any one of which can
be repeated many times to create the series of entries required for cor-
relation purposes. Thus we can have the same stimulus situation and
class of response, measured at the same moment in time on a series of
different organisms of the same class. This is the traditional correlation

RAYMOND B. GATTELL

procedure, e.g., measuring a set of schoolboys on their response to an
intelligence test and then on a mathematics test and correlating the two
series. When carried to a factor analysis it is called R technique. Secondly,
we may correlate over a series of occasions (moments in time) instead
of a series of persons, taking again and again the same set of stimulus-
response (test situation) measures upon one person. This is called P
technique. The three basic designs, or experimental possibilities of cor-

FIG. 1. The covariation chart. From [14].

relation, are called P, R, and T techniques and are shown, with their
transposes, in the Covariation Chart in Fig. 1.

The meaning of the various possibilities in this chart the reader may
work out for himself, or consult fuller accounts [14]. Until the gen-
eralized statement of covariation possibilities was published [14] in
1946, about 99 per cent of all correlation and factoring had been R
technique and the rest Q and P techniques. Since 1946, most of the
theoretical possibilities have been tried in practice and there has been
much wider use of P and Q techniques. It will be observed that each of
the three basic designs, R, P, and T, permits a transposed factoring of
the same score matrix, namely, Q, O, and S, respectively. It is now

Personality Theory from Quantitative Research 269

generally accepted that the same factors are obtainable from any de-
sign and its transpose, so that the decision to use one or the other of
the pair depends upon convenience. For example, with many subjects
and few tests, R technique is appropriate; the converse suggests the Q-
technique transpose.

Furthermore, the three basic designs and their transposes may be
modified further to produce several other useful designs. For example,
incremental R technique, instead of factoring an absolute score for each
person, can enter correlation with the difference of score for each person
between stimulus occasions one and two [28, 102]. Moreover, as sug-
gested above, although factor analysis grew up using naturally occurring,
not experimentally created variation, nothing intrinsic to multivariate de-
sign prevents its also being used with varying stimuli as well as with vary-
ing responses. What has been called the condition response design [23],
which randomizes several controlled, varying stimulus conditions with
respect to one another, is one example of such use. Essentially, it factors
stimuli and responses together, obtaining at once the unitary patterns
of response and their relations to stimulus conditions.

No thorough treatment of the varied possible factor analytic experi-
mental designs is possible here. Our objective in glancing over them is
merely to point out that, as conceived by the psychologist, the factor,
or "source trait" [14], differs from a mere mathematical factor not only
by reappearing in several distinct R-technique studies, as already men-
tioned, but also by its capacity to reappear as the same pattern in these
different experimental designs. For example, a factor labeled Surgency-
and-Desurgency has been found in R-technique analysis, in terms of in-
dividual differences, loading such manifest variables as cheerful, im-
pulsive, talkative, unworried, and some physiological variables, notably
serum cholinesterase concentration. When the same variables are meas-
ured from day to day on a single individual and their trends are inter-
correlated, the factor analysis produces an intraindividual pattern of
just the same form. That is to say, Surgency-Desurgency is a unity in
terms of individual differences and also in terms of function fluctuation
within one person.

Finally, it should be recognized that a more generalized factor
analytic model does not preclude nonparametric variables or functional
relations of factors with variables more complex than those of simple
linearity, as shown by Coombs and Satter [43] and discussed more fully
elsewhere [28]. Until the modified models are developed in terms of com-
putational analysis it will still be necessary to find the factors by the
present linear approximation, operating, therefore, over the small vari-
ance ranges where the approximation better holds. But once the source
traits are recognized and measurable, the nonlinear formulas better qx-

270 RAYMOND B. CATTELL

pressing their relation to any particular criterion can be more accurately
determined in the usual curve-fitting way.

As shown elsewhere [28], approaches in terms of patterns and types
are merely the obverse of the factor approach. The methods suggested
by Horst, Lubin, Meehl, McQuitty, Ellson, Gibson, and others for
finding or using type, pattern, or profile functions are most simply
applied to factors rather than single variables. Indeed there are serious
objections to applying them to variables [28]. A whole new development
of theoretical understanding lies open, in terms of pattern emergent
junctions applied to -factor profiles, once our grasp of the number and
nature of personality factors in man has reached acceptable precision.

THE CONCEPTUAL STATUS AND INTERPRETATION OF FACTORS

The notion that a factor is a single unitary influence underlying many
manifestations rests on the logical premise, found for example in the
writings of the logician J. S. Mill, that covariation betokens a common
cause or elements. It has been suggested, however, [14] that in psy-
chology we should not conceive a unity as an all-or-nothing "reality" but
admit degrees of efficacy or potency. A certain unity may show itself
in R-technique studies with adults but not with children, as some of the
primary abilities do. Another may show itself in all R-technique analyses
but not in P technique, as general body size and general intelligence do,
and so on. In general, a unitary influence is capable of maintaining
its unity only through certain ranges of conditions. Parenthetically, a
unity which shows itself in P technique should in general be expected to
show itself in R technique, but not vice versa; for the levels of variables
as caught at a given moment, in a given population sample, contain both
the fixed individual differences and the internal fluctuation, i.e., they
represent both a trait and a state. Recent evidence agrees in systematically
turning up more factors for the same set of variables in R technique
than in P technique [28].

At this point, we may state that not merely will a unitary influence
show itself as a factor, but that no inference about the existence of
unitary influences is possible, by known scientific method, except through
multivariate analysis, over the range of designs here listed. The existence
of a unity cannot be proved by intuitive perception, or by univariate ex-
periment, or by clinical inference. (For the last is but an approximation
to the statistical multivariate analysis procedures.) However, a con-
trolled experiment may be part of the proof of existence of a unity,
provided that the experiment has multiple dependent variables. For
example, Cureton obtained six R-technique factors in about one hundred
physical-performance variables, and labeled one of these factors ucardio-

Personality Theory -from Quantitative Research 271

vascular efficiency," i.e., capacity to bring oxygen to the tissues. Later
experiments with the same variables in a "high altitude" oxygen decom-
pression chamber, using oxygen pressure as the single independent
variable, showed that it was the particular pattern of variables loaded in
this factor, and no others, which showed deterioration with drop in
oxygen tension. Another example is the discovery of the same particular
stress-response pattern, on the one hand, by Selye's fitting together of
evidence from several kinds of controlled experiment on animals and, on
the other, by our P-technique factoring of the same variables in human
beings [27]. On the other hand the checking of a hypothesis about
unitariness by a hypothetico-deductive sequence of univariate experi-
ments is less satisfactory, because of our inability confidently to integrate
correlational evidence from many different samples.

Granted that a unitary pattern is established, qua pattern — and this
requires statistical checks, such as the salient variable similarity index —
the next step is its interpretation, as a cause or dimension, or at least the
formulation of a progressively testable hypothesis about it. The interpreta-
tion of a factor is sometimes made by considering the set of variables
highly loaded in it, i.e., those whose variance is substantially accounted
for by the factor, and seeking to abstract some quality, content, or
principle common to them all. This is all right as far as it goes, but the
complete procedure requires attention not only to the variables highly
positively and negatively loaded but also to those with essentially zero
loadings. For we infer the nature of a thing not only from what happens
when it is present but also from what happens when it is absent. More-
over, the psychologist needs to be more alert than he generally is to the
possibility that a variable important to his deductions actually was ex-
perimentally included in the researches he is surveying. There are in-
stances of psychologists forming hypotheses on the assumption that a
given variable forms no part of the factor pattern when it was, indeed,
never included in the correlation matrix and so could not possibly
manifest a loading.

The process of deeper interpretation of a manifest factor pattern in
terms of a source trait entity consists usually of a hypothetico-deductive
experimental sequence. Incidentally, it has often been maintained, even
by factor analysts [49], that the multivariate design differs from the
controlled univariate experiment in not being hypothetico-deductive. On
first seeing a factor loading pattern, meeting the conditions of simple
structure, the experimenter forms a hypothesis about the nature of the
source trait. From this he deduces that a previously unused variable
A should be more highly positively loaded than anything he now has
in the matrix, that another, B, should be more negatively loaded, and that
a third, C, should be unaffected. With these three (or more) new

272 RAYMOND B. CATTELL

variables he reenters experiment, to see whether his deduction is con-
firmed.

Incidentally, we should note that this is a more logically exacting,
and frequently a more statistically exacting, test of a hypothesis than
the mere establishment of significant difference on a single variable, as in
univariate experiment. For in the former the experimenter predicts that a
whole pattern of variables will behave in a certain fashion, whereas the
fact that a single variable increases or decreases as predicted, in a uni-
variate experiment, usually leaves inference much more undetermined.
However, it is very rarely that a correct hypothesis for a factor has been
reached in a single act of reasoning, and more commonly we proceed
through a spiral of hypotheses and experiments, gradually raising the
loadings of variables toward that value of unity (when corrected for
attenuation) which permits us to say we have found the underlying
variable which is the factor.

In this connection we should note that though "factor" and "source
trait" are often used as synonyms, yet there is in fact a conceptual
duality. On the one hand, we have the factor (not necessarily a factor in
a single matrix) which is strictly a factor pattern of loadings, as inferred
for a parent population; on the other, we have the concept of a single
underlying "intermediate variable" [75] which causes this pattern. The
pattern is our only means of referring to the source trait, of recognizing
and defining it. (At least, unless there is supplementary controlled ex-
perimental evidence as mentioned above.) And yet we know that this
pattern can never be exactly the same from one sample to another, be-
cause of sampling and experimental error; or from one population to
another, because of systematic influences; or from one technique to
another, since, for example, some variables, which do not vary from per-
son to person, can fluctuate in P technique, over time, and vice versa.
The source trait is the entity, whether it remains abstractly a construct
and concept, or comes to be representable by a literal variable never seen
before; whereas the factor is only a pattern found in some complex
statistical derivatives called loadings.

The identification of the source trait from the pattern can always be
made by understanding and applying the statistical and other laws which
produce the various pattern modifications. But the duality remains, and
must be carefully preserved in thinking. The chief practical reason for
respecting it is that many years may elapse between the recognition of an
invariant., experimentally replicable pattern (including its proof as a
pattern), on the one hand, and its successful interpretation by a correctly
named and conceived source trait on the other. During this period in
limbo, it is important to preserve the pattern with a label which is as far
as possible descriptive rather than interpretive. For the downfall of

Personality Theory from Quantitative Research 273

"faculty psychology" was brought about, not by any fallacy in the
notion of a faculty, but by the fact that the faculties were allowed to
form themselves merely on the patterns of existing words. Incidentally,
the odium which science properly attaches to this verbal vice (even
though the vice is now driven out of personality research where it most
flourished), still attaches itself to some concepts in learning theory and
comparative physiological psychology. Nevertheless, although factor
analysis from the beginning seeks the real evidence of functional co-
variation, instead of unconsciously accepting the false unity of words,
yet the premature attachment of interpretive labels to factors may
prejudice real freedom of thought and experiment. It is for this reason,
and to facilitate work on the establishment of factor patterns per se3
that the present writer has suggested a Universal Index, -with a number
for each pattern believed matched over at least three independent
studies [26, 28]. Some of the factors believed established will be discussed
in the following section.

CLASSIFICATION OF FACTOR PHENOMENA BY MODALITY, DATA,
AND ORDER

Every substantial science has its taxonomy. Each passes through
a phase in which greatest activity is directed to producing order and
stability of nomenclature, before its more comprehensive theories — at
least, genuine comprehensive theories — can hope to emerge. So in per-
sonality study, before "findings" can be discussed in terms of purely
psychological concepts and laws, some statistical and methodological
points have still to be clarified concerning the classification and ordering
of factor patterns per se. It is usual to speak of ability factors and tem-
perament factors, of general and specific factors, of behavior factors and
questionnaire factors, of first- and second-order factprs, and so on. How
correct is it to use these categories, and on what are they founded?
Perhaps four questions will get to the heart of these problems :

1. What is the relation of a factor founded on behavioral phenomena
to one founded on introspective, questionnaire response?

2. How do we know that the factor dimensions we obtain span the
whole personality, or some given domain of it?

3. How do we know when a factor belongs to one modality or
region, e.g., that it is an ability factor rather than a motivation factor?

4. If there are first- and second-order factors, how do we know at
which level we are operating in a given case?

The first two questions need simultaneous discussion. The question
of whether a factor is truly general, i.e., whether it spans the whole
domain of human behavior, involves also asking whether experiment has

274 RAYMOND B. GATTELL

yet covered all human behavior. To ask how we know that a factor is
"general" is in a sense as ridiculous as asking, "How do we know when
the science of physics is finished?" But consider the question, "How do
geographers know when all new land has been found?" and it will be-
come apparent that there may, nevertheless, be possibilities of progres-
sively detailed exploration within a definite, finite area.

Development of an acceptable notion of a total, definitive area of
personality manifestation would have considerable appeal in relation to
several theoretical problems in structured (factor) measurement. In the
first place there are greater difficulties in attempting to integrate a piece-
meal, step-by-step exploration of different areas of variables (as is
feasible in most other scientific areas), compared with those encountered
in an approach attempting to "block in" the main perspectives from a
total realm fixed from the beginning. But a "total realm of phenomena"
requires the concept of a "population of variables," with sampling
properties similar to that of a population of persons. (In terms of R and
Q techniques, or any other pair of transpose techniques, the persons and
the variables have, of course, just such a reciprocal, equivalent relation-
ship.) Variables consist of stimulus-response pairs, so in principle the
possibility exists of defining a total population of stimuli, response habits,
or linkages of these, within a given culture pattern. This special, but
basic, issue is discussed more fully in an appendix to the present article.

Before a solution is suggested here, however, it behooves us to note
that the stimulus-response behavior of the human organism is observed in
three distinct media. It may be observed as behavior, embedded in the
actual life situations, in which case we will call it the life record medium,
or L data. Or it may be observed as introspective responses made to a
questionnaire. This we shall call Q data. A good deal of subtle reasoning
could be followed up about Q data, but the main point is that they
really consist of two distinct kinds of data, with different properties,
according to how the questionnaire is scored. If, on the one hand, we
accept the common meaning of words, i.e., accept the answers as fact
about the individual's consciousness and behavior, we shall call it Q'
data. It yields "mental interiors" [14] and is susceptible to no reliability
coefficient between two equivalent, but different, observers. On the
other hand, if the answers are accepted only as a form of behavior,
i.e., if when S responds "I am shy" we do not take it as evidence of
shyness, but only proof that S so responds, we shall call it Q data. Such
data belongs with the other test approach — objective tests — and so, if we
abandon the introspective, Q' data, we have essentially only two media,
the life situations and the test situations.

Now the notion of a population of variables must rest on the life
situations, for tests can be multiplied according to whim. Lacking re-

Personality Theory from Quantitative Research 275

sources to make a cultural time-sampling of human behavior around
the clock, the present writer suggested using language as a mirror con-
densing this behavior. It was assumed that the dictionary must, by the
twentieth century, have stabilized the number of symbols required to
refer to all aspects of human behavior of interest to man. This symbol
collection is called the personality sphere, envisaged as a finite but un-
bounded set of symbols, which can be represented as points in hyper-
space bearing a spatial relation to one another which is some function
of their meaning relationship.

As a broad strategy of research it has been advocated [14, 28] that
source traits should first be found within a stratified sample of variables
based on this form of L data (behavior in situ, rated with special pre-
cautions). L data should have primacy because (1) according to the
above argument, we can be reasonably certain of covering the principal
dimensions thereby, (2) the factors will appear clothed in terms already
familiar to us (in everyday and clinical language), so that interpretive
hypotheses may be readily reached, and (3) the construction of tests
objectively to measure these primary factors will then be guided by these
hypotheses and will no longer be at the mercy of disproportionate multi-
plication of test behavior merely in some test-convenient areas. Enough
of factor research has followed this strategy to permit development along
the lines indicated.

The question of whether a factor is an ability or a dynamic or a
temperamental trait hats usually been confidently decided, among most
psychologists, by common sense — until those numerous borderline cases
arose which proved common sense inadequate. A special analysis of this
problem [14] has suggested that there are in fact three possible modalities
of factors, as implied above, though any given variable in general
expresses in varying degrees all three modalities. A variable (trait ele-
ment) is defined as dynamic in proportion to the degree that the mean
score (for a population) changes in response to changes in the incentive.
When the score is in an "irrelevance range" of immunity to changes in
incentive, and thus becomes sensitive to changes in complexity of the
situation, the test becomes mainly an ability test. A measure which is
insensitive to both changes in complexity and changes in incentive is de-
fined as a temperament measure. Changes in the situation which are not
changes in incentive are changes in complexity. For logical completeness
this system of definitions now requires an independent definition of an in-
centive. This can be achieved by longitudinal analysis of behavior, de-
fining a goal by consummately responses at which a train of behavior
is found to cease. However, there are complexities in the modality ques-
tion which require such space for discussion that the reader must be
referred to the original [14] statement of theorems.

276 RAYMOND B. CATTELL

Although operational definitions of the three varieties of modality can
thus be obtained, to supplant the rough hunches of psychological
common sense, yet it may be asked whether the modality classification
of factors — on variables — has any value except as an academic exercise.
The answer to this would seem to be that the properties of abilities,
temperament traits, and dynamic traits differ in so many ways that there
is real predictive convenience in having factors as far as possible repre-
senting purely one modality or another. Now the factors that we usually
get from random and unassorted variables and samples will normally
be w holistic factors [14-], i.e., extending across modalities. To take an
even broader example, in a group of children ranging from five to ten
years where physical and mental variables are factored together, we
might get a single growth factor, covering both intelligence and physical
size. On the other hand, in most factor analytic studies we actually get
conditional factors, i.e., factors restricted to a particular modality, be-
cause the variable sample is suitably restricted and all the variables are
in any case presented with certain conditions retained in constancy. For
example, in ability measures a high and sufficient motivation is normally
maintained throughout, whereas, in motivation measures, intellectual
complications as such are implicitly eliminated. Consequently, in what
follows we shall generally deal with (a) ability, (fc) "general personality
and temperament" factors, and (c] purely motivational factors, wherever
"conditional" experiment has been attempted. However, it has to be
admitted that complete modality separation and clarity has not yet
been reached, either theoretically or in the findings.

Our fourth question dealt with the general nature of factors, par-
ticularly the thorny issue of determining whether a factor is of first,
second, or higher order. It might seem sufficient to say that any factor
found by factoring an initial collection of operationally defined variables
is a first-order factor, and that any obtained by factoring the resultant
factors is a second-order factor, and so on. A little reflection will show
that although this should suffice generally, it may fail. By factoring 30
to 50 varied ability tests, Thurstone and others [57, 90], have found
about a dozen primary abilities. The simple structure shows these to be
oblique in relation to one another. When the correlation matrix among
these primary factors is then factored, one or more second-order factors
appear, and that which is most general to all the abilities is considered
to be Spearman's general intelligence factor. It has been shown that its
loadings directly in the tests are the same as those obtained for Spear-
man's general factor, "g" [28]. Thus general ability can be obtained
either as a first-order (primary) factor or as a second-order factor, though
in the first case we have to take special precautions (tetrad differences
made to equal zero) in choosing the variables from which we shall work.

Personality Theory from Quantitative Research 277

Parenthetically one should note that about half the writers on second-
order factors have them wrongly conceived as factors obtained from the
correlations of either (1) the reference vectors, when indeed a cor-
rected inverse of this reference vector matrix is what actually has to be
used, or (2) the factors as literally experimentally measured, by some
battery of constituent subtests. Owing to the immense labor and skill
required accurately to determine the exact hyperplane angles in ( 1 ) as
well as the need to take a mean of several studies, no data fit for a
second-order analysis of personality factors have been available until
quite recently [28], though the present writer must confess to a pre-
mature attempt to determine second-order structure in 1947.

The second-order factors of general anxiety, extraversion, etc., re-
cently found in personality will be mentioned in the ensuing brief survey
of experimental findings, but the important point for the present method-
ological and taxonomic discussion is that in these, as with the older
established general ability factor, it has happened experimentally that
the same factor has, in different settings, been picked up both as a first-
order and a second-order factor. Indeed it is possible to see theoretically
that this could happen. For example, if we started a supposed first-order
ability factorization and happened to use as variables very pure measures
of Thurstone's primaries, and only one measure of each, our factors im-
mediately obtained would be factors generally encountered only as
second-order factors. In personality most measures are not so pure —
the proposed measure for factor A contains also some B, C, etc. Con-
sequently, if there are enough variables, the factoring of these supposed
first-order factors still yields first-order factors, as was found in Lovell's
and Thurstone's reanalysis of Guilford's highly intercorrelating question-
naire measures of factors. But factorings of the Sixteen Personality Factor
Questionnaire, with its relatively pure first-order factor measures, have
yielded second-order factors immediately [28].

Without laboring what may seem an unduly technical point, let us
take it that both theoretically and practically we know that, with rather
unusual circumstances, it is possible to "go right through the floor" di-
rectly from variables to second-order factors without intervening pri-
maries. Consequently one cannot infallibly tell on which floor one has
landed by the merely operational definition that first-order factors are
what you get from test variables and second-order factors are what you
get from factoring factors. Other things being equal, the danger of con-
fusion through going directly to second-order factors, unknowingly, from
variables is much greater when variables are very diverse in nature
and chosen sparsely from a wide area. This argument that factor order
is related to density of variables can be extended, as far as one can see, to
third- and higher-order factors. An acceptable taxonomy and classifica-

278 RAYMOND B. CATTELL

tion of factors according to order, and some other factor concepts, would
therefore be assisted by an operational concept of density of variable
sampling. [For more adequate examination of the assumptions in vari-
able density the reader must be referred elsewhere, 28]. Research has been
able to keep factor orders tolerably clear by a common-sense regard for
frequency of variables in a given area. The accumulating evidence now
strongly suggests that what we commonly call primary personality factors
are on the same level as general intelligence, for "g" appears consistently
as one of them [26]. If this is so, the primary abilities are one order lower
than primary personality factors, whereas the broad, second-order factors
among personality factors are actually on a third level. At least three
factor orders are thus known and used today.

There are accumulating indications that in general the correlations
among first-order factors are smaller than among variables, and those
among second-order smaller than among first-order, so that we shall
probably find that factoring of factors will quickly come to an end,
and probably three or four orders will suffice. Conceptually, the higher-
order factors are organizers among organizers and may carry the in-
vestigator outside the academic field in which he began his work. For
example, the second-order general ability factor might turn out to
be a function of the total number of effective cortical neurones, i.e., a
physiological concept, whereas the primaries are evidently psychological
specializations of a general "relation-perceiving" capacity, in numerical,
verbal, and other fields. On the other hand, the step from one order of
organizers to another may carry us out of psychology in a different di-
rection, into sociology, since one of the second-order factors among per-
sonality factors looks like the orientation of those factors produced by
social status.

In sum, there is a rationale for an initial taxonomy of factors accord-
ing to classification by ( 1 ) medium of observation, covering L, Q, and
T data, (2) modality, covering ability, temperament, and motivation,
(3) order, involving the notion of variable density and the personality
sphere. These are additional to the earlier (4) R-, P-, and T-technique
design origins, and to a further split that will be made later among
motivation factors.

THE PRESENT STATUS OF FINDINGS

The year 1958 is a fortunate one in which to be summarizing actual
findings. Throughout the thirties and forties there was chaos; only in the
last two or three years have the results of many studies finally begun
to fall into place, showing both order and gaps, as the periodic table
did in the generation of Mendel6ef. Since the coherences are still patchy,

Personality Theory from Quantitative Research 279

however, it is desirable to treat each medium and modality on its own,
proceeding to speculative total integration only after the firmer partial
steps have been separately described.

Even in one modality and field of data the decision as to existence
of a confirmed factor pattern rests on several technical instruments. It
requires, for example, first, a test such as Bargmann's [7] showing that
the simple structure rotation position is uniquely significant, and secondly,
the confirmation of the same pattern by at least three quite independent,
blindly rotated experimental studies, all with adequate samples, etc. This
is a matter much assisted by the social organization of research, e.g., the
provision of a universal factor pattern indexing system, as described
earlier, and of precisely defined variables in a master index list of
"markers." Finally, it requires a device to measure the significance
of pattern matching between factors from different studies.

Actually the matching part can rest on three approaches :

1. By establishing a similarity of the loading pattern, on variables
common to the two studies, which exceeds chance expectation by the
usually accepted significances. Actually this similarity of the factors per se
can be examined over more than the loading pattern only, e.g., by a
comparison of their mean variances, the angles to other known factors,
and other properties of the factors, though, in practice, fiducial limits
have so far been worked out only for the loading pattern.

2. By measuring the same population on both factors and showing
that the correlation of the two factors thus measured is not significantly
short of unity.

3. Possibly by the as yet untried hybrid "transformation analysis"
method of Ahmavaara [1].

The first method can be used only when identical variables — marker
variables — are carried through the two studies; the second can be used
either with or without this condition. In either case it will be seen that
the day is past when psychologists might be permitted to match factors
on an intuition that they were "psychologically very similar in nature."
Textbooks are full of factorial castles in Spain built on sincere con-
victions that certain factors confirm the hypothesis set out by an earlier
factor — or even by some verbal definition of the author's concept.

Matching through the pattern in common variables has so far been
done by ( 1 ) correlating loading patterns, ( 2 ) matching patterns by the
pattern similarity coefficient rp which, unlike r, takes level as well as
shape into account [17, 44], (3) using the nonparametric salient variable
similarity index s devised by Cattell and Baggaley [31] specifically for
factor matching. The technicalities of the relative emphasis on these
matching tests cannot be entered upon here, but their recent use on a
series of three to ten planned studies [26, 27, 28] with sufficient common

280 RAYMOND B. CATTELL

marker variables, has shown for the first time, beyond cavil, that simple
structure yields unique, replicable stable factors from study to study.

Fortunately, the detailed documentation needed to substantiate our
account of the 1957 status of replicated factors in L, Q, and T data, and
in general personality and motivational modalities can be omitted be-
cause of the simultaneous publication of an intensive survey of this whole
field [28]. There it is shown that in L data (i.e., life record data using
common verbal definitions of specific behaviors observed in everyday
life) "criterion" situations, some 14 or 15 factors have been established,
each in a minimum of 3 studies. Striking similarities also exist between
some of these and personality dimension concepts, e.g., schizothymia,
anxiety, sex drive, commonly derived from experimental and clinical
fields [101]. The interesting fact is that the list of patterns agrees as
far as the latter concepts go, but that they also go beyond known con-
cepts into dimensions unperceived by the unaided clinical eye. For
example, although the first and largest factor is the "cyclothyme-schizo-
thyme" dimension, long regarded as basic in psychiatry, there is now also
a second schizothyme factor concerned with a pattern of shy withdrawal
(H factor) not associated with hostility, as it is in the first pattern, and
this has not been reliably perceived except by factor analysis.

The familiar clinical concepts of ego strength and superego strength
are now confirmed as independent unities, and it is shown that, in the
normal range, guilt plays a very small part in the functioning of the
latter, in contrast to the pattern perception as biased by clinical sampling.
Other multivariate patterns that can also be recognized from premetric
concepts are dominance-submission, paranoid trend, timidity, and ten-
sion. The surface trait or second-order factor of extraversion-intro version,
as conceived by Jung, is found to resolve itself into at least four func-
tionally independent factors, the most outstanding of which are Sur-
gency-Desurgency and the factors named Parmia and Praxernia. These
three factors are interpreted as representing, respectively, freedom from
past punishment, parasympathetic resistance to threat reactivity, and a
temperamental conversion-hysteria component. All fifteen L-data factors
have been represented in the Universal Index as U.I.(L) 1 through 15,
or in a noncommittal, local laboratory order (of mean variance), by the
letters A through O.

In the questionnaire or Q-data medium, independent factorings and
matchings have similarly established [19, 28, 38, 56, 60], in at least 3
studies, some 18 factors in adults and 12 in children. The most general
of the former are included in the Sixteen Personality Factor Question-
naire, and Thurstone and Guilford-Zimmerman questionnaires; the use
of these against various social, occupational, and clinical criteria has
done much to enrich our practical knowledge and theoretical interpreta-
tion of the factors.

Personality Theory from Quantitative Research 281

The direct impression one gains from the "mental interiors" presented
by the items loaded in these factors is that they agree one to one with
the behavioral exteriors in the L data; but the above objective checks
have to be applied, and in this bridging unfortunately the easier methods
of matching by variables cannot be used, since no variables can be
identical in the two media. Accordingly, cross-media matching has to be
carried out by the second of the above matching methods, i.e., by ob-
taining the two sets of factors on a common population of subjects,
and intercorrelating the factor scores. The results largely confirm the
psychological impressions; such factors as A, E, F, G, H, L, and O run
through both media, showing that a real trait keeps its functional unity
despite different behavioral media, simply changing its dress as the realm
of possible manifestation changes. On the other hand there are some
behavioral, L-data factors — D, J, and K — and some questionnaire
factors — Qi, Q2, Qs, and Q4 — which have not yet been found in the
opposite medium. This may stem from their having much smaller
variance in the other medium, from some real influence associated with
the standpoints of the internal and external observers, or from their
manifestation's being truly confined to one medium.

Since use of the questionnaire and rating techniques, outside the
fully cooperative atmospheres producible in the pure research situation,
is liable to motivational and other distortions, there is urgent practical
need to transfer measurement of personality and motivation factors to
objective tests. By an objective test we mean an exactly reproducible
situation and set of instructions in which the subject's responses are scored
in ways of which the relation to his personality is obscure to him.
(Needless to say, the responses must be understood and scored similarly
by different psychologists.) A considerable variety of tests in the form
of miniature situations, "projective" or misperception tests, stylistic, and
physiological measures — and other forms yet without a name — have been
tried out in factor analytic designs by Brogden [9], Ryans [83], Crutcher
[45], Rethlingshafer [82], Thornton [89], Thurstone [92], the labora-
tory of the present writer, and others. In the last fifteen years, the writer
and his colleagues alone have produced over five hundred different
test designs [15, 21, 42, 30, 24, 39, 36, 37, 26] based on the hypotheses
about individual factors found in the L- and Q-data studies, but the
evidence below suggests that these and such tests as the Rorschach,
Downey, Szondi, etc., still leave important dimensions of personality to be
covered.

Surveys of factors in objective tests were made by the present
writer in 1946 [14] and by French in 1953 [55]. However, a firm
evaluation became possible only with Bargmann's devisal of a significance
test for simple structure [7], with the introduction of the salient variable
similarity index for factor matching [31], and above all, with the fruition

282 RAYMOND B. GATTELL

of a long-term research plan designed specifically to carry systematically
representative markers through several independent population samples,
factor extractions, and rotations. The matching of seven studies has
now proved to be good; in 1955, there appeared a final integration and
interpretation, revealing twelve factors of a relatively high degree of
definition and invariance and six of a less satisfactory degree. These
eighteen factors included [26] those found by Eysenck [50], Thurstone
[92], Gruen [37], and Dubin [36] on special groups and showed that
these factor patterns persist with little change through normal and ab-
normal, younger and older populations.

On the other hand, the alignment of these T-data factors with those
in L and Q data is far from simple. Three studies [40] have been carried
out, using the second (and only possible) method of matching. They
show that there is good matching of L and Q factors, but that relatively
few of the L and Q factors have yet been located in objective tests, and
that in some cases what appears as a first-order factor in T data is sec-
ond-order in the other realms. For example, the anxiety factor [U.I. (T)
24] in objective tests correlates substantially with the distinct question-
naire anxiety factors O, Q4, L, and C(-), now known to form a second-
order Q factor, while the invia-exvia factor [U.L(T) 32] appears to be
a second-order factor among a group of three questionnaire factors
defining introversion-extraversion behavior. The L- and Q-data factors
which appear to be most directly represented in objective test factor
equivalents are: G, superego, M, Autia, I, Premsia, K, Comention,
N, shrewdness, and L, Pretension, or paranoid tendency. Quite apart
from matching with other media, the psychological meaning and con-
sistency of the T factors are good, and one can recognize among them
such factors as general character development, anxiety, assertion, psy-
choticism, general inhibition or restraint, neuroticism (checked by
Eysenck by criterion rotation on neurotic groups), hypomanic tendency,
"corticalertia," and a superego-like set of responses called "Critical
Practicality" or U.L(T) 19.

Although the primary research task at present is the confirmation of
the patterns themselves in further age groups, cultures, etc., and the dis-
covery of test designs that will measure them with increased construct
validity, their use in applied psychology, e.g., against clinical, occupa-
tional, and educational criteria, would also greatly help the task of inter-
pretation. Essentially, we have reached the vital point where the factors
are verified as patterns, and prior to which speculative hypothesis forma-
tion woud have been a waste of time. But the stage is now set for more
intensive hypothetico-deductive experiment, and much of this can be
carried out by smaller-scale, univariate designs, eliminating the complex
multivariate methods necessary at the "blocking in" phase.

Personality Theory from Quantitative Research 283

Some special objective test factoring developments should be indi-
cated in order that the full scope of present results may be properly per-
ceived. These developments diverge from the line of general tests for
general personality factors in that they factor a series of tests all involving
the same kind of response, notably musical preference reactions, esthetic
tastes, humor preferences, and motivation responses. The last deserves a
special section later. The factoring of laughter response to jokes is based
on the Freudian theory that strength of reaction to wit betrays the
strength of particular repressed tendencies in the unconscious. Stable fac-
tors, not quite as clean-cut in simple structure as for general objective
tests, have been found in these realms by Andrews [5], Eysenck [49],
Gattell and Anderson [30], and others. Except for a few special cases,
e.g., Eysenck's relation of the "bright, clear color," picture preference
factor to Surgency-Desurgency (hysteria-dysthymia) and CattelPs
definite relation of the "sexual and debonair wit" factor to L and Q
factor H, Parmia, the relation of these factors to those in other realms of
expression remains to be determined.

All the matching and confirmation problems so far considered have
been those among the different media using R technique, in which the
great majority of published factor analyses are expressed. In the last ten
years, however, a brief but vitally interesting collection of P-technique
analyses has arisen, i.e., of longitudinal factor analyses within single in-
dividuals, and these have planfully used the same variables as those in
the R-technique factorings [34, 35, 27]. In cross matching from R to P
no exact statistic can be used, because sampling is not comparable, and
certain systematic differences would be expected in the pattern from the
same source traits by the two situations. However, it is notable that con-
siderable agreement exists, both in L and R data, and that the factors
A, C, E, F, G, and H in L data, and U.I.(T) 17, 19, 21, 22, 23, 26,
and 29 are believed now to be found both in terms of individual differ-
ences and in patterns of diurnal or other fluctuation (function fluctua-
tion) within one individual.

P technique lends itself well to investigation of psycho-physiological
connections, because many physiological measures fluctuate appreciably
and are yet of a nature which permits of their being repeatedly tested,
without the disturbances resulting from repeated application of psycho-
logical tests. This work, summarized elsewhere [27], has led to more pre-
cise delineation of sympathetic, parasympathetic, and stress reactions,
and to recognition of the physiological associates connected with swings
in the major psychological trait patterns. Notably here is the identifica-
tion of high serum cholinesterase and low alkalinity of saliva with states
of surgency, of reduced metabolism with the paranoid states, and of
higher blood pressure, pulse rate, and ketosteroid output with anxiety.

284

RAYMOND B. CATTELL

P technique has also been extremely useful in discovering the factor
structure of motivation manifestations, but since special theoretical de-
velopments are necessary in presenting the motivation factor findings,
they are deferred for separate description. The rapid review above,
which should be supplemented by reading in the systematic factual sur-
veys now available [28, 56], shows that the harvest of consistent, con-
firmed empirical findings has relatively suddenly become far more ex-
tensive than is commonly realized. This is especially true of certain ex-
perimental and applied fields, which could avail now themselves of the
new structure with great advantage.

COMPLEX FUNCTION AND CONFIGURAL, TYPE PREDICTION
FROM SOURCE TRAITS

At this point, with actual factor findings available for illustration and
testing of the assumed properties in the model, we can return to a fur-
ther development of the theoretical position set out under Personality Re-
search in Relation to the Two Basic Scientific Methods and The Logic
of Factor Analytic Experiment above. This explicit treatment is nec-
essary not only to develop the full use of personality factor measurements
of all kinds, but also to lay the foundations for the more complex re-
search concepts encountered in factoring dynamic motivational data.
Principally we are now concerned to make a more explicit statement of
the assumptions in the mathematical model which we are using, and to
make a further transformation of some current rather vague psychologi-
cal concepts in personality into exact operations related to our model.

Two major assumptions (besides homoscedasticity) are made in fac-
tor analysis:

1. That linear relations exist (a] among the variables (so that prod-
uct-moment correlations can be used) and (b) between the factors and
the criteria,

2. That factor functions, whether of the first or higher powers, are
influences which combine additively, rather than by some more complex
interaction.

The linear and the additive statements are constantly confused in
discussion. As to the correctness of the first it can be said that real cur-
vilinear relations have very occasionally been found in personality vari-
ables, but that in the typical correlation matrix used, as many as two
or three thousand plots have sometimes been examined without finding
a single significant departure from linearity.1 On the other hand, there

*A practical comment on this situation was made by Flanagan: "extensive
study of large samples in research during the war failed completely in establish-
ing anticipated curvilinear relationships" [55].

Personality Theory from Quantitative Research 285

are indications, though scarcely proofs, that curvilinear relations between
factors and criteria are fairly common. For example, ratings on occu-
pational proficiency suggest, but do not yet prove, that middle values in
certain temperament factors are sometimes more effective than either
extreme value.

If there were definite demonstration of a relation of this kind be-
tween a factor and a criterion — instances are Eysenck's suggestion that
neurosis may be more common with high and low than average in-
telligence, and the present writer's finding that low ego strength (C
factor) is found with both high and low extremes of rigidity [11, 24] —
it could be handled either by modifying the specification equation, or in
applied work, simply by scoring the factor on a new scale from a new
zero point. In the former case, a parabolic curve would be represented,
expressible, according to its axis, by making the criterion C a function
of plus or minus P2 or F%. Of course, this modified specification equation
could no longer strictly be used as an integral part of ordinary models for
factor extraction and rotation. However, it is likely that by choosing
small ranges, in which the linear approximation is good, the current model
can continue to be used to find factors, even when curvilinearity or non-
additivity ("joint functional relationship") exists. Then, in using the
factor, over wider ranges of variation, the appropriate nonlinear function
could be found and used. For the speculative nonlinear factor models of
Coombs and Satter [43], though of great interest and promise, have not
yet been worked out in a way that would permit extraction of factors,
and specification equations, from experimentally observed relations
among variables.

The second assumption — that factor functions are additive — could
break down in many different ways. It might, for example, be necessary
to change the equation to product relations, or to include both simple
summation and product terms (interaction terms in the analysis of
variance conceptualization), or to introduce products involving higher
powers of the factors. The variations are indeed infinite. The notion so
frequently raised — but rarely in clear or testable form — in clinical psy-
chology, that the profile or pattern may have effects through its shape,
independent of the effects of the absolute levels, is another way of bring-
ing up the same question.

The normal procedure in science is to adopt the simpler model unless
and until results prove that a more complex one is required. Although
clinicians, in particular, have brought up apparent examples requiring
a more complex model, the present writer knows of no well-substantiated,
cross-validated example that cannot be worked out as well by the simpler
model. Empirical results are conflicting. Improved results are claimed for
configural scoring by Meehl [76], Saunders [84], Fiedler [53], and

286 RAYMOND B. CATTELL

Lubin and Osburn [70], but the simple linear treatment was found
superior to pattern, configural, or complex function methods by Tucker
[94], Bell [8], Ward [97], Lubin [69], and Lee [68]. However, it is very
•probable that true cases of more complex, interactive configural effects of
factors exist and we shall return to the problem in a moment, after a
clearer development of assumptions in the main model.

Practically, it can be said that as far as locating factors is concerned
— as distinct from using them in more complex situations — we can either
(a) locate them initially among variables (the majority) in which the
relations are linear, or (b) take small ranges, as stated above, in which
the linear approximation is close enough to permit the model to work.
In this connection it should be pointed out that artificial examples by
Thurstone [90], Bargmann [6], and others, in which complex functions
(higher powers of factors and products of factors) have deliberately been
introduced as the basis for correlation, have nevertheless always proved
susceptible to factor analytic reduction, and the complex relation-
ship has been found to appear in terms of its nearest simple additive
equivalent.

In any case, a great deal of work remains to be done in psychology
with the present proved effective model. As usual, armchair speculation is
running far ahead — or astray — of effective integration of theory with
actual research. Before speculating indefinitely and philosophically, it
behooves us to understand fully the implications of our present model
and to use it as a tool to advance psychological knowledge (and there-
fore knowledge of the required model modifications). These implications
are:

1. That although conceptually we analyze the individual into
dimensions, any of his acts is an act of the total personality. We repre-
sent this integration by giving influence to the majority of dimensions
(in the specification equation) in estimating the magnitude of each
response.

2. Since loadings can be both positive and negative, we recognize
that some factors help in some circumstances and interfere in others.
As we shall see (under The Evidence for Motivational and Organic
Factors; Erg and Sentiments) in the special case of motivation factors,
this difference of sign is interpreted as evidence of conflict.

3. The same level of response can be reached, according to the
specification equation, by persons having different (but equivalent)
factor endowment patterns. This equivalence of different behavior can
readily be perceived as true in psychological observation, and we then
say that the "quality" of the performance is different in the two cases,
even though the quantitative level is the same.

4. Although the factors added are all in standard equivalent scores,

Personality Theory from Quantitative Research 287

these scores are not identical in terms of any absolute dimensions and
qualitative characters of the units from factor to factor. For example,
we have no evidence that the variability (range) of people in intelligence
is smaller, equal to, or greater than their range in Surgency. But in the
specification equation we add ability, temperament, and motivation
units, or habit strength and frequency units, in the same realm of standard
scores.

5. The /s, or factor loadings in the specification equation for a par-
ticular stimulus-response variable, may be considered as the psychological
dimensions of the situation.

6. This last statement, like most quantitative psychological state-
ments, has meaning only relative to a given population. Indeed, the
factor pattern itself, similarly, is something defined in terms of a popula-
tion (or in P technique, a population of occasions in the individual's
life).

7. The usual factor measurement assigned to an individual is only
a statement that he is at the given level on that factor at the moment of
measurement. At what level he will be on other occasions is to be inferred
from our psychological knowledge of the factor, the statistical findings on
function fluctuation (complement of stability coefficient), and the gen-
eral psychological laws of learning and maturation for that factor. For
example, the factoring of dynamic data reveals a sentiment pattern of
"interest in one's profession" affecting a whole pattern of interest and
skills. In industrial psychology it is not unusual to predict a person's
future adjustment to a particular occupation from an "occupational
interest blank" measuring his interest before he is actually in the occupa-
tion. The learning from repeated actual exposures to the occupational
situation is likely to increase the strength of this factor to a point at
which individual differences are likely to have little relation to those
before learning. Consequently the effective use of factor source trait
measures requires general psychological understanding of the way in
which maturational and learning laws are likely to affect their future
course. Indeed, one of the major superiorities of source trait formula-
tion over mere use of variables, in applied and experimental psychology,
is the fact that these meaningful unities can be effectively brought into
relation with general psychological and physiological laws of growth and
learning.

With this brief statement of the psychological implications of the
present factor model, let us turn to its relation to configural and
typological prediction. The recent furor of enthusiasm for the latter
seems to have confused, rather than developed, two basic truths:

1. That configural methods resolve either into (a) use of complex
mathematical functions of the profile of factor (or variable) scores or

288 RAYMOND B. CATTELL

(b) a simple recognition of types, together with an Aristotelian logic,
"This is a dog: therefore, it may bite/' i.e., the use of memory rather
than calculation. Ellson, McQuitty, and Lubin and Osbum are among
the few who have recognized that the task in the latter approach is
simply to key a species against the criterion properties of the species,
without any immediate attempt to "understand" the property in
terms of general scientific parameters defining the species. Conversely,
Horst [64] has demonstrated that many attempts at configural pat-
tern scoring are properly examples of — and would be more clearly
conceived as — a modified specification equation, as in configural method
(<z), using the regular factor or other parameters in mathematical
functions.

2. No matter which way one decides to use types, they can be found
as modal patterns in a distribution of patterns in a space of dimensions
or parameters common to all types. Their final separation, however, may
require addition of dimensions, for particular pairs or sets of types, not
common to all types. Thus raises the basic proposition that "trait" and
"type" descriptions are not in different worlols, but are simply reciprocal,
complementary, and mutually dependent ways of analyzing and abstract-
ing the same data. This can be seen most clearly in the case of R and Q
techniques, which are mathematical transposes. In other words, traits
(or, beyond psychology, attributes) are abstractions made from cor-
relating variables over sets of organisms, and types are abstractions made
from correlating organisms over sets of traits. The approaches duplicate
in statistics the division in language between adjectives on the one hand
and nouns on the other (or, if processes rather than persons are our
concern, between adverbs and verbs) .

The present writer has explored elsewhere [28] the implications of
the two brief statements above, at the much greater length which alone
makes possible intelligible, if not final, formulation of the problems. At
the risk of apparent dogmatism, the following points from that discussion
may briefly complete the present picture :

1. Regardless of the mode of further use of types (la or Ib above),
the discovery of types as modes (2 above) inevitably falls into two dif-
ferent approaches with different conceptual systems :

<2. One may take complex functions of the elements in the profile,
e.g., a function best predicting the criterion, and find the modes
in this univariate (complex function) distribution.

b. One may measure the resemblance of every individual to every
other, by some pattern index operating upon attributes, and find
the pattern modes ("correlation" clusters, in terms of the index
values) among people.

Personality Theory from Quantitative Research 289

2. The latter process has so far been used in a way which could not
yield unambiguous results because :

a. The indices employed, e.g., the correlation coefficient in Stephen-
son's O' technique,2 or in d2 advocated by Osgood [80] and Cron-
bach [44], are not truly comparable from study to study, or they
emphasize "shape," "level," or "deviation" in the profile match-
ing at the expense of the total similarity. A pattern similarity co-
efficient rp has been proposed [17] which takes all three aspects
of a pattern profile into account and which has resulted in good
functional grouping of national culture patterns [20], Maha-
lanobis's general solution is also valuable [see 81] here. Paren-
thetically, it should be mentioned that discriminant function
methods are of no use for typing; they require one to have some
prior means of designating criterion groups and thus are circular
in argument.

b. Variables, instead of independent factors, have frequently been
used as elements in the profile. Since variables may be highly cor-
related, one area of behavior may then be weighted out of all
proportion to another, i.e., any figure for the similarity of two
people is purely arbitrary, depending on the variables thrown into
the matrix. The problem of sampling variables remains to be
solved. Getting profile similarities with factor measures as elements
solves this to the extent that whatever is represented is equally
represented, though it still leaves the question of whether un-
known regions of behavior are being omitted.

A grouping of persons in occupations, according to similarity of
profiles of personality factors, has recently been attempted by Day and
Meeland [see 28] and it seems that thereby more invariant groupings
are obtained, as in the national culture pattern studies [17, 20], than in,
say, McQuitty's [74] use of patterns on variables (test items). However,
it is not only the advantage of statistical invariance but also of psy-
chological meaning which points to handling patterns and types in terms
of factor elements, as a more intelligent strategy.

Ideally, taxonomic and predictive problems are best handled in an
integrated combination of type and parameter methods:

1. Choose variables on which all types can be measured, factor, and
determine by rp on factor profiles the modes (types) and their positions
in this framework of generic transtype dimensions.

2 Q' is best used instead of Q, because the three primary factor analytic designs
and their transposes have been symbolized as R and Q, P and O, T and S.
Locating types, on the other hand, involves only finding clusters in the correlation
matrix and thus stops short of any true factor analysis, and is best indicated by Q'.

290 RAYMOND B. CATTELL

2. Factor within each type on dimensions both common to types and
peculiar to the within-type variance of that type.

3. Handle any given individual by assigning him to a type, accord-
ing to f 1 ) , and then predicting the deviation of his behavior from the
mean behavior of that type in terms of his endowments on the within-
type factors.

This approach allows for the existence of differences of behavior be-
tween types which are categorical, Aristotelian, and not yet predictable
from the parametric traits by any pattern emergent function known
to us.

As Thorn dike, McQuitty, Ellson, and others [74], have shown for
normal persons, types, as pattern modes, are mostly found in occupational
skills and social role behaviors, rather than in basic personality source
traits, which tend to be normally distributed. In abnormal persons,
however, as Wittenborn's data tend to show [101], we may be dealing
with segregating, modal patterns; and such patterns are clearly evident
in some biological abnormalities, e.g., phenylketonuria, Huntingdon's
chorea. Nevertheless, even in basic source traits, one might expect
distinct types to emerge, if samples from different races and cultures are
included in the analyzed sample. Currently in progress, is an experiment
to determine the constancy of personality factor patterns across seven
different countries, but regardless of the degree of constancy found, it
should be possible to determine transcultural factors by factoring a
sample with one representative from, say, each of a hundred cultures,
and then plotting the distribution of patterns on these factors in a larger
sample taking many from each country.

The problems of complex factor function, configural and type pre-
diction are complicated. We have the mathematical and statistical tools
for handling them, but we can use those took intelligently only when we
attend to what is rather than what might be. Conceivably some con-
figurations will give "emergents" — in Lloyd Morgan's and Bergson's
sense — which cannot be predicted by any mathematical combination or
discoverable function of elements, and then a sheer type approach must
be used; but it is to be hoped that the above stated combination of type
and trait formulation, with its greater intelligibility and scientific appeal
by generalizability, will fit the findings.

THE EVIDENCE FOR MOTIVATIONAL AND DYNAMIC FACTORS;

ERGS AND SENTIMENTS

With the foundations of our model thus further clarified, we can turn
to a new realm of psychological application in which some more exacting
demands are made on it. The application of multivariate experiment to

Personality Theory from Quantitative Research 291

dynamic motivational areas is quite recent — since 1948, in fact — but has
had its results so quickly confirmed and has led to such a promising de-
velopment of dynamic calculus, that it deserves special theoretical at-
tention— and also special practical attention from the clinician.

As indicated in the earlier discussion of modality, conditional factors,
largely peculiar to one modality, can be obtained by restricting variables
to that modality. Thus, in this dynamic field the variables must be un-
questionably motivational The plan has been to redefine "attitude"
as a basic motivational surface manifestation3 and, by factoring on a
foundation of attitudes, to explore the dynamic structure of personality
in terms of drives, sentiments, self -structures, or whatever other forms
may turn up among the attitude elements. An attitude is a stimulus-
response habit, expressible in the paradigm :

"In these circumstances ... I ... want so much to do this with that."
Stimulus situation Organism Response (defined as a course of

action: "to do this"; of given in-
tensity: "wants so much"; gener-
ally involving some reference to an
object: "with that.")

It is supposed that every major dynamic system must eventually ex-
press itself in attitudes, and in the courses of action that go therewith, so
that by experimental analysis of these it should be possible to reveal
the underlying systems. Parenthetically it must be stressed that the above
definition and measurement of individual attitudes cannot be equated
with much of the attitude measurement that has been done in sociology,
because ( 1 ) an attitude here is not narrowly conceived as "for or against
an object" but is free to assume any of a wide range of emotional
qualities, e.g., curiosity about, or anxiety about, an object. (2) The
self-conscious, self-evaluation, verbal, opinionaire method of measuring
attitudes is not accepted as valid. Instead, a group of diverse (physiologi-
cal, learning, perceptive, indirect verbal) sub tests of motivation strength
is used. The traditional verbal opinionaire correlates only about .3 with
the pool of general motivation measures and thus deals with only some
peculiar one-tenth of the total attitude strength variance.

In developing the new objective measures of attitude strength, over
fifty widely chosen subtest devices [21, 32, 35] were intercorrelated (with
respect to each of a number of representative attitudes). Factoring these
devices of measurement methods revealed that all the tests by which
strength of a motive is supposed to manifest itself do not "go together."

3 Other terms are unsuitable. "Motive" may mean either a structure, as we in-
tend, an incentive, or a process. "Interest55 is equally uncertain, commonly meaning
interest in an object rather than in a course of action.

292 RAYMOND B. CATTELL

There is not a single motive strength underlying all, but no fewer than
five independent factors. These have been called motivation component
factors to distinguish them from dynamic factors, discussed later, which
structure the whole area of attitudes. Although the names may not
appeal to experimentalists, the fact is that the character of these five
factors so closely corresponds to the psychoanalytic concept of id, ego,
superego, and -unconscious complex components (plus a physiological
component) that we have tentatively so named them. For example, the
motivation component factor we have called the "id" contains all the "I
want" manifestations, together with high fluency on good consequences,
low fluency on bad, autistic misbelief and misperception phenomena,
rationalization, and other ego-defense mechanisms [32], At the same
time it has no loading in the manifestations found in the realistic ego
component, such as knowledge and skills in reaching the goal of the at-
titude, readiness to make effort and to learn, tendency to relate cogni-
tively to other interests, etc. ; and it lacks the GSR response, blood pres-
sure changes, and other "complex indicators" present in what we have
called the unconscious complex component [32].

Later it was shown that these primary motivation factors could be
resolved into one or two second-order factors permitting, in the first
case, a single over-all measurement of integrated motivation strength
for any given attitude. A second phase of research next developed in
which an objective test battery, to cover the main second-order factor
or factors, on the above foundation of evidence, was applied to each
of 30 to 50 variously chosen attitudes, in order to factor the dynamic
structure, Le., to find whatever drive or acquired dynamic4 habit patterns
exist among human interests. In comparing the outcome with others,
e.g., the work of Torr [92] and of Guilford and coworkers [see 28], it
should be stressed that each attitude, though represented by a verbal
statement, is actually measured by perhaps 30 or 40 responses made in
objective tests, i.e., in the battery of GSR, word association, fluency, etc.,
measures as just described, and validated also against objective criteria,
notably the actual amount of time and money spent on a given attitude-
interest. The intercorrelating and factoring of 30 to 50 varied, but
objectively measured, attitudes, on 3 substantial samples of young adults,
has shown remarkable agreement of outcome [21, 33]. The results
indicate that most dynamic structure factors are drive patterns, but
some others correspond to socially acquired patterns of attitudes which
may be called sentiments, e.g., religious, career, patriotic, sports and
games, hunting, mechanical, etc.

4 We have refrained from using the generic term "habit," for dynamic structures
generally, because many psychologists rightly include in habits many purely
cognitive patterns and motor skills which are at the service of any dynamic
structure.

Personality Theory -from Quantitative Research 293

To avoid entanglement in the prolonged verbal, nonoperational dis-
putes about instincts, drives, and propensities, e.g., those of Watson,
Murray, and McDougall, the new term ergs has been suggested
specifically for the patterns found in factoring of motivational traits
which do not correspond to any known sociocultural institution (as the
sentiments do) and which closely resemble in emotional and goal quality
the drives seen in the primates and the higher mammals. Incidentally,
Anderson [3], Haverland [63], and others have factored motivation
manifestations in the rat and have arrived at similar identification of

Attitude level

Sentiment level

Ergic level

FIG. 2. The dynamic lattice. From [20].

factors with drives. Their work suggests that measures of drive strength
in most univariate learning experiments with rats, etc., have been of a low
order of accuracy and could be improved by representing the drive by its
factor loaded variables instead of a single variable. For example, "period
of deprivation of food" does not load the hunger drive factor any better
than "degree of restlessness," or as well as "speed of running to previous
food goal." Incidentally, in earlier qualitative observations Harlow [61]
had already suggested that the experimentalist's faith in hours of depriva-
tion as a good operational measure of hunger strength is ill-founded. The
factor studies suggest it has a validity of only about .5 and that a com-
posite battery, with appropriate factor weightings, would give a much
improved hunger-tension measure.

294

RAYMOND B. GATTELL

The ergic patterns which seem best substantiated in man are sex, self-
assertion, escape, fear (or anxiety), parental protectiveness, gregarious-
ness, rest-seeking (sleep), curiosity, exploration narcistic sex, appeal,
construction (two studies only) .

Each of these can be scored from responses on six to twelve attitudes
saliently loaded therein — and let it be reiterated that each of these at-
titude response strengths is itself determined, not from any single verbal
assertion, but from some four objective subtests (e.g., GSR, autistic mis-
belief, word association, information) covering about forty actual re-
sponse measures.

A valuable check exists in the finding that the same dynamic factors
have also appeared in P-technique study [35], wherein a clinical case
was tested on the same attitude strengths from day to day for eighty
days. Factoring of occasion-to-occasion variance again brought out a
simple structure In wrhich such drives as sex, fear, parental protective-
ness, etc., appeared. In this case a factor score could be assigned to each
erg for each occasion, and comparison of these tension levels with the
diary and clinical records showed that the strengths of the drives from
day to day can be closely connected with recorded stimulus situations
and deprivations, thus providing evidence of the ergic nature in-
dependent of that inferred from attitude content.

In both R- and P-technique studies the constituent attitudes were
carefully chosen to provide a check on the ergic and sentiment hypoth-
eses. For some pairs were chosen to have a common goal character
but quite different sociocultural content and history, whereas others
were all concerned with a particular social molding influence, e.g.,
religion, career, sports, but were clinically considered to exercise very
different kinds of drive satisfaction. This design arose from the con-
ception of the dynamic lattice, according to which attitudes are organized
in learned, environmentally determined subsidiations, forming chains
(crisscrossing in lattice formation) from the most recently acquired and
culturally complex at the distal (left) boundary to the given, biological
consurnmatory goal activities at the proximal (right) border. The con-
ception of the dynamic lattice is both clinical and experimental in origin.
The notion of "subsidiation,35 derived from Murray [78], arises largely
from clinical experience with free association, in which superficial in-
terests are followed to deeper and deeper drive goal interests. But the
notion is also rooted in experiments on animal learning, in which be-
havior Z', leading to goal Z, is followed by further learning of behavior
Y', leading to goal Y, when the situation is made such that the animal
finds it cannot immediately start from subgoal Y.

Essentially the dynamic lattice concept is at only a low level of
Abstraction from the facts — it is an undeniable, almost literal description

Personality Theory from Quantitative Research 295

of the way dynamic habit systems get organized in any organism that
must learn ways to its drive goals. But, with the aid of the hydraulic
model, and other specific postulates about its mode of operation, the
dynamic lattice develops into a powerful model which has already per-
mitted a range of important hypotheses to be more exactly investigated
than hitherto. For example, according to this view of attitude structure,
we should expect that all attitudes which subsidiate to a particular ergic
goal would wax and wane in strength simultaneously with changes in
that goal need, and would thus appear in correlation studies as loaded
in a single factor. This involves the assumption also of conditions most
quickly defined as "the hydraulic analogy" in which, from rate of flow
observed at certain "outlets," the underlying "feed-pipe" connections can
be inferred from observed covariations in the rates of flow.

For variations in drive tension level, the above argument is clear
enough and works as expected, but the corresponding argument for the
appearance of a sentiment structure as a factor pattern is beset by more
qualifying assumptions. Indeed, the initial failure [21] to find sentiment
patterns suggested possible flaws in the argument, but later findings
support the main position. If all of a set of n attitudes are involved in a
particular sentiment, then it will follow from the unity of the social in-
stitution involved that the individual who has most frequent occasion
to express himself through one of these will also be in a position more
frequently to express himself through the others. If frequency of rein-
forcement has influence on the strength of a habit, then all the attitude-
habits socially involved in a single sentiment will tend to be simul-
taneously weak or simultaneously strong. Consequently we should expect
to recognize such connected habits by their being loaded in a single
factor. Actually, in the last resort, a distinction must be drawn between
two kinds of sentiment structure, namely, sentiment as an object of inter-
section of attitudes and sentiment as a subgoal, with somewhat different
covariance properties; but this refinement must be left for better dis-
cussion elsewhere [28].

The finding that these sentiment patterns have variance much lower
than that of the ergs, in some studies, but quite comparable variance in
others, can be most intelligibly connected with the type of objective at-
titude measurement used. When the test measures are predominantly
those of the first motivation factor,5 notably the defense mechanism and

5 For clarity of discussion it is necessary to stress again here the distinction made
earlier between the five or more factors found in motivational measurement
devices, on the one hand — which we shall call motivation component factors —
and the fifteen or more factors found among attitudes — which we shall call
dynamic structure factors. The first classification has to do with the more basic
sources of energy, whereas the latter deals with general dynamic structure.

296 RAYMOND B. CATTELL

autism devices which caused us to identify this first factor with the id
component in attitudes, ergs stand out strongly, but sentiments do not.
On the other hand, there are at least indications that when the measures
are those most highly loaded in the ego component, the sentiment factors
become of substantial variance. This is what would be expected, since
the id is mainly concerned with desires, whereas the ego is built of habits
adjusted to reality and derived largely from social learning. Indeed,
several such necessary connections between dynamic factors and motiva-
tion factors (see note 5 above) are deducible from the hypotheses and
urgently need investigation as a check on the whole notion of cross
classification of dynamic manifestations by motivational and dynamic
abstraction.

Meanwhile, it is necessary to make those controlled experimental and
background-associating checks (in ways suggested in the next section) on
the ergic and sentiment interpretations which are so strongly suggested
initially by the content and selective patterns of these factors. Recently,
Humphreys and Lawrence made possible a first check on the nature
of the dynamic factors by correlating them with the general personality
factors as measured in Q data (the Sixteen Personality Factor Test).
Although a few significant correlations were found, suggesting some real
second-order connections between temperament and drive patterns (the
sentiment patterns significantly had few associations), yet the results
clearly confirmed the general contention that these are "conditional"
[14] dynamic factors in a realm new and distinct from that of the estab-
lished general personality factors.

THE DYNAMIC CALCULUS OF ERGIC STRENGTHS, THE SELF-
SENTIMENT, CONFLICT, AND INTEGRATION

The findings outlined in the previous section have opened up what
amount to new systems of dynamic concepts and of dynamic calculations
which have considerable importance for theoretical developments in psy-
chology, as well as for clinical practice, though it may take some time
before the latter is realized. This system, which must be examined in the
present section largely in hypothetical terms, because of the scarcity
and recency of experiment, includes the following concepts: the dynamic
lattice, a dynamic vector calculus for ergs and sentiments, a formula for
drive tension, definition and measurement of conflict and integration,
analysis of the self-sentiment and its defenses.

As to the dynamic lattice, its initial rationale and sources for fuller
development have already been indicated [28], To summarize briefly:
the whole dynamic structure of the individual, conscious and unconscious,
can be expressed in this lattice, and the relations of attitudes, sentiment

Personality Theory from Quantitative Research 297

structures (including the self-sentiment), and ultimate subsidiations to
ergic goals can be graphically represented. This lattice representation
must also admit feedbacks (or reverberator}' circuits). Ergic and senti-
ment structures therein can be located by factor analysis, and it is
probable that other relations therein can be handled systematically by
models successful in hydraulics and by the mathematics of lattices and
networks. It is further hypothesized that individual differences in certain
general properties of the lattice as a whole will relate to other personality
characteristics; notably, that the amount of long-circuiting (summation
of goal distances in a representative sample of attitudes) will determine
the individual level of general anxiety [U.I.(T) 24], that the com-
plexity (count of cross connections and feedbacks) will relate to strength
of personality integration (factor Qs), and that the function fluctuation
of attitude strengths (occasion to occasion) will correlate (negatively)
with the factor of ego strength [U.I.(L) 3]. The truth of the last has
been independently shown by Cattell [24] and Das [47].

Turning next to the proposal for a dynamic vector calculus we en-
counter the following theorems. First, a vector summation of the at-
titudes loaded in an ergic factor will give the tension level of that erg ( 1 )
in terms of individual differences, as the general level of need in the in-
dividual in the given life-situation, and ( 2 ) using occasion measures, for
a particular occasion in a given individual. In the former case we can-
not tell, from the factor measure alone, how much the ergic tension
measure expresses (a) a congenitally greater need in that individual
for that drive satisfaction, (b) a (temporary) greater stimulation of that
drive by the particular stimulus situation in the life environment, or (c)
a level dictated by lesser opportunities for goal satisfaction, with the same
amount of stimulation. More completely analyzed, the measurements in-
volved in the hypotheses of ergic tension level may be expressed in the
following equation:

E = S[C + H + (P - aG}] - bG

where E is the ergic tension as measured by the factor score, based on
the motivation measurement devices, S is the stimulation given by the
existing life situation to the erg, C is a constitutional component in need
strength, H is a component from the previous history of exercise of the
drive as a whole (including, e.g., any repression of it), P is a physio-
logical condition (temporary) component, G is the extent to which the
drive is receiving satisfaction in the general life situation, and a and b are
constants representing the effect of the last directly on physiological
satiation and on psychological satiation respectively. All this analysis is
initially at the level of hypothesis from general psychological observation,
but it leads to more precise experimental testing through the new ability

298 RAYMOND B. CATTELL

to measure drive tension In man. In verbal terms, the expression
S[C + H + (P — aG\] is drive strength, I.e., the tension level measured
apart from the satisfaction incurred. Further, the part within the square
brackets is need strength, i.e., the strength apart from stimulation. A
recent P-technique study [35], showing that the function fluctuation of
£ is of about the same order as individual difference variance and that
it relates closely to known environmental stimuli, suggests that we shall
find the larger part of the variance in E to lie in S. In any case, the
possibility now of measuring E, instead of some single variable alleged to
represent it, makes a new level of computational accuracy possible in
motivation and learning experiments using such a formula. It will be
observed that the principal difference from current learning theory for-
mulations is the use of sums instead of products.

As to the calculus of sentiments, the following operations become
possible. Since any set of measured attitudes, being themselves vectors,
can be added vectorially to a single resultant fas used by the engineer
in a polygon of forces), it should be possible, from inspection of all the
attitudes in the dynamic lattice gaining satisfaction through a par-
ticular object, to calculate the ergic projections of a single vector which
will represent the strength (and quality) of the (object-intersection)
sentiment as a whole. Suppose now, we accept the preliminary findings
that there is no erg of pugnacity, but that the strength of anger-destruc-
tion behavior is simply a function of the total strength of the ergs
frustrated by the removal of the object of a given sentiment. Then we
have a possible experimental check on the above calculation; namely,
that the strength of anger-destruction ("aggression") behavior at the
threat of removal of the sentiment object concerned should equal the
figure calculated for the attitude resultant. Incidentally, the only opera-
tional sense that can be given to the "for and against," sociological, habit
of talking about attitudes is through considering that the concept really
applies to sentiments rather than attitudes. Then "for" represents a bal-
ance of satisfactions from the continued existence of the object, w^hereas
"against" means that the various attitudes intersecting in the object (not
represented in the lattice) sum to a negative total and would thus give a
gain of satisfaction if the object were done away with.

This calculation of the amount and kind of ergic satisfaction in an
object has especial value in attempts by social psychologists to anticipate
the new equilibrium Hkely to be reached when one institution is abolished
and another substituted. For vector addition of ergic tensions in attitudes
can be carried out not only within one individual but, with suitable
attention to metric, in any dynamic system, e.g., a set of sentiments in
one individual or a single social attitude rooted in many individuals. In
particular it has been proposed that calculations on group morale, using

Personality Theory from Quantitative Research 299

the concept of .group synergy [41], be made by obtaining vector re-
sultants of individual attitudes "I want to continue to belong to this
group" [16] summed over all the individuals in the group. This di-
rection of development raises many radiating problems of definition
which unfortunately cannot be followed up here. For example, it raises
the question whether what we measure in ergic tension is the analogue
of force or of energy, and wiiether ergs should properly be given equal
weight in the specification equation, as they are by the usual use of
standard scores. For the present, until more research is done in genetic
and physiological components fC and P) by multiple variance analysis
designs [28], the ergic tension breakdown is the least secure of the for-
mulations. We are safer to say that our proof is simply of an erg as a
dynamic factor, such that any one of an array of stimuli excites it and
any one of an array of responses reduces its excitement. The form
of these patterns of possible stimulation and response is presumably
culturally determined but the degree to which a person possesses them is
partly genetically determined.

If now, as supposed above, the ergic factors are unities of tension
level, whereas the sentiment factors are unities of experience of repeated
reward, the adaptation of the factor specification equation to a peculiarly
dynamic analysis equation must be expressed by having two kinds of
factors, which we can call E, or ergic tension, factors, and M, or engram,
factors, where an engram means any kind of empirically, factorially
demonstrable unity due to patterns of experience — and therefore cover-
ing sentiments (object intersection and subgoal) and complexes, as far as
present psychological conceptions go. Thus we have :

(omitting, for simplicity the nth terms and the specifics) where Ru is the
magnitude of response in a given attitude situation j of the individual i3
and the £"s and AT s are fs endowments in the ergic tension and engram
learning levels. Again it will be observed that, in contrast to some
prominent learning theory formulations, this starts out with the simpler
assumption of summation rather than multiplication of drive and ex-
perience components. But this is not basically important, for summation
is an approximation to multiplication, and in neither field is experiment
yet exact enough to decide.

Turning now to the fourth concept listed earlier, we come to the
derivation, within this dynamic calculus, of formulas for the degree of
conflict and of integration (or adjustment) in a particular dynamic
system or a person. This development begins with the interesting induc-
tive conclusion that in factoring dynamic variables the loadings, un-

300 RAYMOND B. CATTELL

like those found for general personality factors, tend to be predominantly
positive. On reflection it is easy to see that this must be so, because no
attitudes, sentiments, drives, or individuals could continue without a
balance of satisfaction over dissatisfaction. When an attitude (habitual
course of action) is negatively loaded on certain ergs and positively on
others it means that a necessary adjustment has been reached in which
the individual denies himself satisfaction on one drive (by suppression,
repression, or any mechanism capable of producing adjustment) in order
to gain greater satisfaction on another. Conflict can be regarded either
as a transient state of indecision (in which case it is either a conflict of
means to an agreed ergic goal or a conflict of ergic goals, as Maier,
Brown, and Farber, and especially Mowrer, Masserman, and Maslow
have brought out) or as an accepted compromise, in which one erg
continues to get satisfaction at the cost of greater or lesser dissatisfaction
to another. The present studies [16, 18, 21, 32, 33, 35, 42] in so far as
they deal with settled attitudes, are concerned with conditions in the
second phase of "fixated conflict," rather than the first phase of "active
conflict"; but the conflict is not less real because it has ceased to be
conscious and the focus of decisions.

Our proposition is, therefore, that fixated conflict in any attitude,
erg, or sentiment system is shown by the existence of opposite sign factor
loadings. Consequently the amount of conflict, for any of the possible
referents (attitude, erg, sentiment, person, group), can be obtained
by calculating the amount of cancellation which occurs, i.e., the sum
of negative values, or the ratio of this to the arithmetic sum.

The concept of integration or adjustment is honored more in fine
phrases than in calculations; when calculations have been made, as by
Hartshorne and May [62], Hull [66], McOuitty [73], Das [47], and
others, they generally turn out to score something other than dynamic
integration, e.g., conformity to the group, central tendency of profile,
stability of attitudes, agreement only of self-ideal and self-concept, etc.
If we accept the definition of dynamic integration as the extent to which
one dynamic trend does not undo another [20], in other words, that it
is the ratio of total satisfaction to total drive need (considered in a stable
situation over a sufficient length of time), then a true calculation of in-
dividual dynamic integration is possible. Taking a stratified sample of
important life attitudes, and performing a P-technique factorization on
the individuals to be compared, we should obtain standard motivation
factors for all, but with loadings on the representative attitudes differing
for each individual. The expression for adjustment would then be the
total algebraic sum of each such person's attitude factor matrix divided
by the total arithmetical sum thereon. Questions of metric must be
handled, and particularly, the problem of perhaps substituting for that

Personality Theory from Quantitative Research 301

weighting of the factors given by the latent root values a weighting on
some "absolute energy of a drive" concept. However, these refinements
do not invalidate the main conception and their treatment must be
deferred to a less condensed presentation. The above can be summarized
as:

c— "

and /=1-c

Where C is conflict, / is integration, and s is factor loading. A test of this
drive measure conflict formulation has recently been made by Williams
[100] who found the predicted highly significant relation (average
-f- = 0.6) between C and (1) patient-nonpatient difference, (2) the
ego weakness factor, and (3) psychiatric evaluations of conflict.

In the fifth and last concept above, namely, the self-sentiment, we
encounter one of the most difficult conceptual problems in the whole of
psychology. How does one bring the self and the self-sentiment into the
dynamic lattice? Most writers on the self-concept — Sherif and Cantril,
Rogers, McDougall, and the psychoanalysts among them — assign to it a
powerful dynamic influence in controlling impulses and view it as a
comprehensive clearinghouse to which all kinds of behavior systems
are referred. To state the conclusion of much discussion, the self-concept
must be considered as central in a widely ramifying sentiment which
subsidiates to almost all satisfactions, but particularly those of security
and self-assertion. It does so because foresight concerning the physical,
social, and moral preservation of the self is actually a prerequisite for
the satisfaction of most other drives and sentiments.

By such reasoning the self-sentiment should appear in the dynamic
lattice as a late development, i.e., most distal from the ergic goals, and
affecting a wide range of attitudes but especially those directed to
social reputation, self-control, and the general preservation of the self.
Such a single broad factor, over and above the sentiment factors con-
cerned with career, hobbies, etc., has now been replicated in three in-
dependent researches [21, 33, 35]. In the P-technique study [35] it was
possible to check that the loadings of attitudes in this sentiment agreed
with the emotional values in which the individual (and his self-concept)
had been raised. But much remains to be investigated in these terms, and
indeed, we are only on the threshold of measurement and calculation
and its relation to clinical background data.

In connection with the ego one must also raise the question of what
quantitative multivariate research has contributed to the knowledge of
ego defenses, hitherto based on clinical insights. To an appreciable de-
gree, the extension of motivation study into this area is linked to the
meaning of so-called projective tests. As suggested elsewhere [4] "pro-

302 RAYMOND B. GATTELL

jective" has been an unfortunate term, implying a definite projection
process in a realm of behavior where research has been so poor that no
knowledge of the real processes at work has yet been established. In a
new approach [18, 28, 42, 981 it has been suggested:

L That this class of tests be defined as mis perception tests, indicating
that the essential operation is one of measuring the deviation of a per-
ception from a norm, or from a measurable reality.

2. That the assumption that this is the result of a single dynamic
tendency, "projection/3 is wrong. Wenig's factorization [42] of a variety
of misperception and defense mechanisms revealed five distinct factors:

(a] poor or incorrect cognitive furniture (low "g" and information),

(b) naive projection, (c) true projection, (d) fantasy, and (e) autism.
Thus some misperception is due to ego-defense dynamisms and some to
processes of a different kind, but since we are concerned with misper-
ception phenomena only in so far as they throw light on dynamisms,
we shall here follow up only the former. Further research on the same
lines might well prove additional defense dynamisms, but for the present
we have proof of functional independence (in terms of individual dif-
ferences) of naive projection, true projection, rationalization, and prob-
ably, reaction formation, and identification.

3. Clinicians have seldom stated, still less established, whether mis-
perception should be positively correlated with conscious self-integrated
needs or unconscious rejected needs. They generally seem to assume that
if a person sees more aggression in a TAT picture he himself has more
than average aggression, and the same is vaguely indicated for any other
trait Actually, the foregoing proof of independently acting defense
mechanisms shows that they can both reinforce and oppose one another
in the direction of the misperception resulting from one and the same
given dynamic source in the person. So long as the test is not designed to
separate their interfering actions, only a poor correlation could be ex-
pected— and only poor correlations are in fact found. When objectively
scorable misperception tests are used, the evidence points to a low positive
correlation between misperception and overt behavior. This can be
reconciled with the fact that positive correlations are also obtained with
unconscious motivation — as shown in the Cattell and Baggaley "id
factor" [32] — in varied misperception tests only by the additional hy-
pothesis of covert-overt proportionality [4]. This supposes that those per-
sons who, through constitution (or through infantile experience), deviate
initially from the norm, will be culturally pressed, in overt behavior,
toward the norm but tend not to reach it. Thus the internal (repressive-
suppressive) adjustment results in the covert, unconscious component
deviating on the same side of the norm as does the overt component (not
oppositely, as is so frequently and unquestioningly assumed). Paren-

Personality Theory from Quantitative Research 303

thetically, this seems to apply as much to nondynamic, temperamental
dimensions as to drives and is supported by quite independent evidence
from genetic studies with the multiple variance analysis method, showing
a marked predominance of negative interactions between hereditary and
environmental variances [28].

With the general personality dimensions measurable with tolerable
reliability, the ergs and engrains measurable by objective test batteries,
and the demonstration of location of defense mechanisms by multivariate
methods, the stage has at last been set for reliable experimental investiga-
tion of fairly complex hypotheses. Already, in the last two years, some
very provocative relations have been found in terms of second-order
factors among dynamic traits [28], and of significant correlations between
the defense mechanisms and certain personality traits [42, 99], as well
as relations between the strength of the self-sentiment and of the superego
factor and anxiety level.

These correlations and factorings are leading to more exact under-
standing of the dynamics of the self, but although the general concepts
in this area continue to be in approximate agreement with the non-
quantitative and necessarily vaguer clinical concepts, it cannot be said
that the multivariate methods have yet unearthed anything correspond-
ing to, or positively requiring, the psychoanalytic concepts of conscious
and unconscious. The defense mechanism findings just discussed offer
fragmentary evidence which would inferentially fit the hypothesis of the
unconscious (but also others) ; and there are manipulative experiments
on forgetting, conflict, projection, etc. — few but well known — which
point the same way [50, 71, 77, 85, 86, 88]. There is also the evidence
interpreted as showing id, ego, and superego motivation components in
any attitude [32], as discussed above. But on the whole, the division
found between the two second-order factors in motivation components
[32] is better described as that between integrated and reality-tested
systems on the one hand, and wishful, reality-distorting (but not neces-
sarily unconscious) systems on the other.

SUMMARY AND SYSTEMATIC ANALYSIS OF THE PRESENT

The development of a neat abstract formulation of concepts and
postulates in regard to the theories of personality growing from multi-
variate quantitative approaches was impossible in the early stages of this
presentation because terms did not exist in common language to handle
them. With the preceding survey of the empirical findings, of the de-
pendence of constructs upon procedures, and of the unique logical char-
acter of the concepts developed in this area, a more compact formulation
can now be made. In this summary we shall conclude by relating con-

304 RAYMOND B. CATTELL

cepts to the issues posed in the editorial outline, as promised at the be-
ginning, and shall add reference to some additional purely theoretical
developments not reached in the survey of experiment.

Of the concepts which have grown uniquely in this field and which
the reader needs to keep in mind for the neater formulations, he can be
reminded by a glossary of such terms as source and surface trait, simple
structure, L, Q, and T data, specification equation, cooperative factor,
transcultural factor, pattern similarity coefficient, transposed factor anal-
yses, incremental R technique, P technique, Surgency, ego strength,
Parmia, motivation component factors, dynamic structure factors, ergs,
engrains, subgoal sentiment, distal end of lattice, ergic tension level, need
strength, self-sentiment, naive and true projection, misperception meas-
ure, and the law of overt-covert proportionality.

The first part of our more basic summary will simply condense the
main exposition, reviewing conclusions in an order which can nowr be a
compromise between that necessary to show historical sequence in re-
search and that desirable in terms of logical dependence and clarity.
After a 16-point precis, we shall turn to the editorial discussion outline.
We shall then summarize the bearing of the present paper on each of the
editorially suggested items, giving special attention to those not already
directly treated.

1. Personality research on a quantitative basis proceeds both by the
classical univariate controlled experiment of the older sciences and by
the multivariate analysis designs which have been developed in the life
sciences; but at the present phase of personality research, good strategy
would give the latter much larger scope to define the functional unities
with which controlled experiment may best concern itself.

2. The definition of the factor analytic model merely begins with the
matrix transformation theorems, i.e., reduction of many vectors to few
coordinates, as known to the mathematician. Its scientific use and
rationale involve many more restrictive conditions and more complex
ideas, notably, use of statistical criteria of unique rotation, planned and
statistically tested matching in cross validations from research to research,
coordinated experiment to examine the degree of scientific "efficacy" of
factors in R, P3 and R-incremental designs, and the checking and
further interpretation of factors through their use in controlled ex-
periment. However, all uses have in common the aim of finding naturally
occurring, underlying functional unities in stimulus-response variables.
Underlying unities of pattern, too complex to be perceived by unaided
clinical or univariate experiment, thus constitute the "intervening
variables" around which concepts can profitably be developed, and upon
the interactions among which more basic laws can be hopefully built
than upon innumerable empirical paired-variable relations.

Personality Theory from Quantitative Research 305

The specification equation, which is the central theme of the model,
has the following basic form :

Rii = SjiFu = Sj,F2i + • • • + SjnFni + SjFfi (1)

and defines the stimulus situation ; for the response R by a pattern of
"situational indices" or dimensions, ^i, Sj*, etc., and the organism also
by a set of dimensions, Fi, F*, etc. It is thus a more developed stimulus-
response formulation.

3. A distinction must be drawn between the concept of the factor
pattern, by which factors are first recognized, and that of the under-
lying source trait, which is expected to manifest itself in modified pat-
terns according to sampling and other transformations. The interpreta-
tion of factors is a hypothetico-deductive spiral employing both factor
analysis (contrasting the nature of variables with high and low loadings)
and controlled experiment; but multivariate experimental designs exist
which permit the same sequential, causal inference as in manipulative
univariate experiment, and with higher powers of definition.

4. Factors may be classified (a] according to the three exclusive
sources of observation, as L-, Q-, or T-data factors, (fc) according to
modality of variables, as ability, temperament, and dynamic factors, and
(c) according to density of variable representation as first-, second-, and
higher-order factors. A sampling of behavior space is implied in the
concept of the personality sphere.

5. At present there is acceptable replication and confirmation of
about 20 ability factors, 30 general personality factors, and about 15
dynamic factors. In personality manifestations it is assumed that the same
real dimension will express itself in all three media, but so far, research
has not succeeded in finding many factors crossing all media. Cross-
media checking cannot be carried out by the "s" index [31] because
of absence of common markers but must be by direct correlation or
transformation analysis. Some five second-order factors have been found
in each realm, and some of these, such as the anxiety factor among L-
and Q-data factors, and inhibition factor among dynamic traits, give
substance to concepts long appearing as clinical hypotheses.

6. The model of the specification equation assumes linearity of var-
iables to factors and additive relation among factors with respect to the
criterion, i.e., no "joint functional relationships" in the mathematicians'
definition. Demonstrable absence of fit is rare, but certainly exists. How-
ever, in artificial examples it has been shown that factor analysis is able
to yield the correct factors, but with linear approximations to more
complex relations, when complex relations do exist. The model admits
the possibility of different individuals' obtaining the same score in dif-

306 RAYMOND B. CATTELL

ferent ways, and in general fits psychological conceptions of the per-
sonality as an integrated set of traits.

7. In the absence of computing methods to fit the speculative models
of Coombs and Satter "43], or of parametric analysis of the predictive
device of Lubin and Osbum [69, 70], or to accommodate to the whole
range of possible mathematical functions, the best approach is to isolate
factors by the present model but to determine the curves of their relation
to various dependent variables by controlled experiment. Thus, one can
arrive at more complex functions than the simple specification equation,
in the rather uncommon instances where the latter breaks down.

So-called configural scoring is but a special case of predicting criteria
through a general mathematical function of elements, i.e., use of the
developed specification equation. But types, as pattern modes, are of two
kinds — special purpose, based on distribution of a complex function,
and general purpose, based on Q' technique using the pattern similarity
coefficient. Trait and type approaches are face and obverse of the same
method, however, and are best used in conjunction, types being defined
as modal patterns in profiles based on factors as elements.

8. Elementary dynamic variables, defined as attitudes, have been
found to factor, i.e., to give reproducible simple structure patterns, as
readily as ability and other modalities used earlier, and indeed, to be
particularly responsive to P-technique designs. Reasoning from the na-
ture of the experimental design, and the relation of discovered factors
to stimuli, etc., strongly suggests that the fifteen or so replicable patterns
found are those of drives, specifically defined here as nine ergs, and some
six engram (mainly sentiment) patterns. Engrams are learned patterns,
resident in memory. The measured strength of such a pattern in an in-
dividual corresponds to the degree of exposure (frequency-reward learn-
ing) to the social institutions through which the component attitude-
habits are learned. The specification equation for an individual attitude,
which resolves a symptom into dynamic factors, amounts to a quantita-
tive "psychoanalysis" of motive.

9. An alternative analytic split of dynamic data can be considered,
in which the forms of manifestation of motive (for any attitude) such as
misperception ("projection53), ego defenses, learning, attention, physio-
logical and autonomic response are factored for a single attitude. This
constitutes a complementary or reciprocal treatment to 8, for instead of
factoring a single operational manifestation of many attitudes it factors
many motivational manifestations of a single attitude. It has yielded
some five ego-defense mechanisms and four or five "motivation com-
ponent" factors, which seem to correspond to id, ego, and superego in-
terest components, present in every attitude. Although these are "cross
factorizations," certain systematic relations would be expected between

Personality Theory from Quantitative Research 307

the motivational level factors and the dynamic structure factors, notably
that the "ego component55 (realistic and realized habit expression of an
interest) should be stronger in engrains than in ergs.

10. The dynamic lattice of subsidiated attitudes, which is a construct
almost at the descriptive level, together with the hydraulic model used
to make predictions in it, leads to positive designs for analyzing dynamic
structure both by multivariate and manipulative univariate experiment.
As indicated in 9 the multivariate method has successfully abstracted
from the lattice both ergic structures and engram structures. Conse-
quently, the typical expression for the strength of interest in the course
of action defined by an attitude is a weighted sum of ergic tension levels
and engram (sentiment, or Freudian "complex," experiential) com-
ponents, as follows:

(7y or)6 R,- = sjelEi + ' ' • + sjenEn + • • •

+ SjmiMi + ' ' ' + SjmnMn + SjMj (2)

where E and M are respectively erg and engram factor scores and Mj
is an engram absolutely specific to the given attitude.

An attitude is thus a vector, amenable to vector summation to get the
ergic and engram composition of any dynamic system in an individual,
or in group phenomena, from the interaction of many individuals.

By hypothesis the ergic tension factor levels can be broken down as
follows:

E = S[C + H + (P - aG)} - bG (3)

i.e., drive strength S[C + H + (P — aG)], need strength [C + H +
(P — &G)], situational stimulation level S, satisfaction or satiation levels
aG + bG, constitutional and historical components C and H in the need
strength itself, as well as a physiologically manipulable physiological
component in need strength P. The relations of these formulations to
learning theory formulations have been commented upon. The princi-
pal differences are use of additive instead of product relations, and the
splitting of both drive and reinforcement experience components into a
pattern of dimensions instead of a single term. The stimulus situation is
also expressed as a pattern of dimensions.

11. Central in the notion of total dynamic personality structure is the
empirically demonstrated self-sentiment structure, as well as the super-
ego and certain independently functioning defense mechanisms. These
agree approximately with the clinical concepts but take on more definite

6/j or strength of interest in a course of action, if the response is inhibited,
but RJ if we literally measure the magnitude of response in the given course of
action.

308 RAYMOND B. GATTELL

properties and implications. For example, the self-sentiment is distal in
the dynamic lattice and is correlated significantly with certain drive
strengths and general personality dimensions. Further, the two second-
order factors in motivation components suggest, instead of the clinically
prominent division into conscious and unconscious, a division into inte-
grated, reality-oriented dynamic systems and wishful, unrealized, not-
reality-tested but not wholly unconscious systems.

12. The notion of ergic vector projections permits a dynamic calcu-
lus of interest investments and conflict, by R technique for the average
man or P technique for a particular clinical case. Fixated, if not active,
conflict can then be measured as the sum of negative projections in a
dynamic system, divided by the sum of positive projections. A meaning-
ful value, / = 1 — C, can be derived for I, the adjustment (integration)
of an individual, where C is the conflict index, equal to S^( — )/S,r( + ),
the s's being ergic projections for a stratified sample of important every-
day life attitudes. This index has been shown to be substantially cor-
related with clinical ratings of adjustment and with patient-nonpatient
differences [100].

Conceivably, by applying the calculus of interest strength in atti-
tudes, as in Eq. (2) above, one can give a meaning to psychological
energy, through multiplying this force (interest strength //) by a meas-
ure of distance achieved toward a goal. This speculative notion is intro-
duced to indicate that a considerable possibility of further theoretical
development resides in the present formulations.

13. Each use of the factor analytic specification equation and the
included source trait measurements, in personality, supposes only that
the relation holds at the given moment 3 in terms of the factor measures
also taken at the moment. However, the approach through factors
(rather than specific tests, etc., as in most current applied psychology)
implies, and opens up, the experimental possibility of supplementing
statistical prediction (using only strictly factor analytic, actuarial, "in-
stantaneous35 estimates ) by use of general psychological laws dealing with
the expected change of factors and stimulus situations over time. Indeed,
with the replicable factors now available, it at last becomes possible to
proceed to these laws of growth, learning, physiological determination,
etc., of factor strength which will integrate present personality theory
with other general psychological laws. Similarly, the definition of stimu-
lus situations by situational indices, i.e., by assigning scores of common
psychological dimensions to all stimulus situations, opens up possibilities
of generalization about stimulus situations which should make possible a
psychophysical calculus permitting extension of behavioral prediction to
situations that have not yet actually been used to determine a specifica-
tion equation.

Personality Theory from Quantitative Research 309

14. The metric of factors, like that of most other psychological
scales, is essentially ordinal. However, because of the fact that many
component subtest measures enter into any factor scale there is a better
argument than usual for the assumption that the distribution is a
Gaussian one and that equal intervals might be found by cutting accord-
ing to units giving a normal distribution. In the case of P technique, and
of interest measurement, it has been proposed [28] that ipsative rather
than normative scoring should be used, and most have agreed with this.

Whatever the nature (not the metric) of the units for the single
factors — in a sense analogous to the dimensionality of the units of physics
— the dimensionality of the units of the variables predicted by the
specification equation must be multiple. For we add at least three
modalities — abilities, temperament traits, and dynamic traits (the last
being dual) — when, after factor analysis, we put the individual to-
gether again in the single emergent behavior defined by the specification
equation.

15. Although the factor analytic model has been considered by most
psychologists only as a means of measuring individuals (as to their traits
or states), it should theoretically be equally important in defining and
measuring stimulus situations, their dimensions and changes. (In terms
of a vector for each situation S = si, #2, /ss, . . . , sn.) The definition
is not physical but psychological, i.e., in terms of the behavior of the
species of organism reacting to the physical world. The relation of the
psychological valences of a situation, thus factor analytically determined,
to the physical properties constitutes a considerable new area of possible
development of psychophysics, beyond the purely cognitive psychophysics
of the Weber-Fechner tradition, into an affective-dynamic psychophysics
peculiar to each culture and every species of organism.

16. There exist certain forms of behavior and of temporary change of
behavior, notably that under the sociological conception of "adopting a
role," but also under "mood change," which in terms of the model could
be expressed in either of two ways: (a) by changing the situational in-
dices, the /s, corresponding to the verbal equivalent that "the person per-
ceives the situations differently" or "the situation has changed its mean-
ing"; (b) by changing the quantitative terms for the individual's person-
ality (the T"s, or E's or Afs), but by some formulation which indicates
that it is a temporary phase.

The latter seems preferable, for much current talk about the "new
look in perception" makes the basic theoretical error of introducing
unnecessary terms. Instead of introducing "change of perceptual mean-
ing in the situation" as a middle term, one may simply say that, in a role,
the behavior changes and that the terms for personality have changed. Of
course, in some kinds of perceptual change it may actually be more

310 RAYMOND B. CATTELL

economical, for a number of predictions, to introduce a term corre-
sponding to the introspective experience of a change in perception, and
this alternative we have considered under a below, illustrated in the case
of a role change producing a change of perception. In any case, in role
change and most other changes of perception, it does not suffice to
depend on introspection, as sociologists and psychologists in perception
have done, for knowing when the change exists. It must be inferred from
a change of pattern in behavior. This is best determined, at least in the
case of roles, by applying the profile similarity coefficient, rp, to the be-
havior of many people in many situations, using Q' technique [23, 28].
Granted the demonstration of a role change, it can be formulated in
our model as stated:

a. As a perceptual change, through introducing a second type of
situational index Sr multiplying or otherwise modifying the usual
nonrole /s, in such a way that the new response is accounted for,
thus:

Rjri = SrjSjiTu + ' ' ' + SrfrnTni (4)

the Sr being determined by factoring role behavior.

b. As a temporary change in personality, definable for the role by
introducing a profile of trait modifiers, tn to tm, which can be
applied as a grid, the same for all people, as they step into the
role, thus:

Rjri = SjifajTu + • ' ' + Sjn(trn)Tni (5)

Incidentally, 4 should remind the reader that our general formulation
of situational indices is such that every stimulus situation is really con-
ceived as a Chinese "nest of boxes," situation within situation, which
for initial simplicity, we divide into an immediate situation s and a life
situation S.

Let us now examine the series of 12 editorial rubrics to see where
supplementation of the previous 16 points is necessary.

Point 41}, Background Factors and Orienting Attitudes, was dealt
with at the beginning. On point {2), the Structure of the System, in
terms of independent, intervening, and dependent variables, we had to
run off the rails of traditional analysis, because factors do not fit into this
univariate scheme but require other modes of thinking. Unless time
sequences are introduced into factor analytic experiment, only incom-
plete inferences are possible as to whether the factor is a dependent or
an independent variable. But in either role it has the advantage that it
directly represents a systematic rather than an empirical variable, i.e.,
it permits direct and sure reference of the dependent-independent rela-

Personality Theory from Quantitative Research 311

tionship to concepts or constructs, more certainly than in nonpositivist
approaches, and at the same time gives comprehensive information about
the intervening variable that is missing in positivist approaches. This
essential point will be evident from considering summary 2 above and the
earlier, more detailed exposition which it covers.

Editorial point {3>, the Initial Evidential Grounds for Considering
the Formulation Promising, is, in the case of this system, scattered
widely over psychology, in the success in relating test variables to ability
concepts in education, etc., and in the various empirical findings system-
atically organized here.

As to point {4>, The Construction of Function Forms, it has been
seen that the specification is initially restricted to linear relationships
between variable and variable, and factor and variable, but that the ap-
proximation, when nonlinear, is usually good enough to permit factors
to emerge, after which all manner of complex functions and factor
profile derivatives can be determined by controlled experiment. Prac-
tically all the theoretically assumed mathematical function relationships
in ordinary R technique, in incremental R technique, in condition-re-
sponse factoring, in P technique, in factor matching, etc., have been
tried out experimentally and found to check with the formulations as
well as, or better than, those in any other area of mathematical formula-
tion in psychology. On the other hand, the more recent formulations for
change of perception of a situation, change of personality in a role, the
origin of ergic tension, and the measurement of internal conflict have not
been checked to anything like the same level of certainty. Point -6K on
Mensuration Procedures, is handled by 14 in our summary above.

Editorial point •{6}, concerning the Formal Organization of the Sys-
tem, is answered in practically every page of our presentation. It will by
now be realized that this system has a high degree of formal organization.
Yet one should note that in the main, such organization has not been ar-
rived at by clearly formulating axioms or postulates and proceeding to
check inferences. Instead, there has been much groping and intuition,
and especially, the formulation of limited-scope subsystems to fit particu-
lar areas, before attempting any more general or "grandiose" postulates.
The main ultimate axioms are (a} personality or the totality of behavior
can be analyzed into a number of functional unities or factors, (b) that
these interact additively (in the first approximation) to produce the
degree of behavior observed, (c) hierarchies can be found among these
factors such that each primary affects only a limited area of behavior,
but higher-order factors organize several primaries. This implies that
factors are not uncorrelated but demonstrate their independence by in-
fluencing independent sets of variables.

Point O}, the Scope or Range of Application, has been sufficiently

312 RAYMOND B. GATTELL

illustrated, as also has I8>, the History of the System to Date in Mediat-
ing Research. Potentially the scope includes the Investigation and expla-
nation (at a certain level) of all multivariate phenomena, psychological
and nonpsychological, e.g., meterological, biological, physiological. For
although factor analysis and related multivariate methods were born in
psychology, they belong to the life sciences and social sciences generally,
constituting a second main approach not needed in the simpler realm
of the physical sciences. In the last five years there has been a remarkable
growth of factor analytic findings in physiology, biology, economics,
sociology, and anthropology. For example, Sokal, Stroud, and others
have also used it in the complex taxonomic problems of entomology,
Damarin in physiology, and Driver in anthropological culture-pattern
study. These extensions are to be welcomed by the psychologist, for there
are statistical and logical problems in the method that will be far more
readily solved when factors receive wider scientific exemplification and
when diverse statistical developments from these new sources are inte-
grated with it.

Within psychology the chief interrelations, present and potential, are
with learning theory, whenever learning theory begins to deal with
motivation effects in a more positive and detailed fashion, and especially
when it progresses from means-end learning to integration learning.
Secondly, there are models developed from the present system, but set
out elsewhere [28, 41], which deal with matters of increasing importance
to social psychology; viz., the degree of constancy of personality factors
across cultures, the theory7 of common scales where factor patterns are
not identical, the means of relating personality measurements to measure-
ments of the behavior of groups per se [41], and the dimensions of
culture patterns.

Thirdly, there are major possibilities in relation to genetics and
physiology. In the latter it has been shown that autonomic and stress
states can be factor analytically identified and integrated into a total
"trait and state" formulation of individual behavior [28]. In the former
there have been explorations, independently by W. Thompson and the
present writer, of the possibility of understanding gene structure more
specifically by combinations of factor analysis and existing biometric
genetics.

Concerning editorial point <9>, the Extent of Supporting Evidence
and of Evidence Embarrassing to the System, the chief instances of the
latter arise from (a] occasional evidence of curvilinearity, the handling of
which, experimentally and in terms of slight modification of the model,
has already been discussed, (6) a few clear instances of the "permissive
relation," in which one factor refuses to come into operation at all until
another reaches a certain level, (c) the state of confusion which still

Personality Theory from Quantitative Research 313

persists in the "intermedia" realm [40] after the initial discover}* that
the same personality factors do not neatly crop out simultaneously in all
three media, (c) is partly resolved (1) by finding that first orders (pri-
maries) in one medium are, in at least four instances, second orders in
another, e.g., in the case of anxiety and exvia-invia, and (2) by the good
line-up found in general between rating and questionnaire media. Only
the problem of separating "instrument factors,53 and of pursuing further
questionnaire-objective test matchings, still remains. What concerns in-
completeness of model and method, rather than sheer incompatibility of
evidence, will, however, be dealt with separately below.

The Applicability of the System to Areas beyond That in Which It Has
Been Used, editorial point -CIO}, has been indicated for areas outside
psychology under editorial point <7> above, but its concepts and prin-
ciples have frequently been urged as powerful aids in psychological
contexts other than personality theory:

1. In determining the dimensions of syntality in groups, as a pre-
liminary to relating them to structural arrangements and population
means.

2. In clinical psychology, the factor analysis of the individual case
by P technique could give more positive understanding of the in-
dividual dynamic lattice and lead to more precise dynamic laws.

3. In learning experiments, it is contended that there would be a
better chance of hitting upon laws of a systematic nature if drive
strengths on the one hand, and learning effects on the other, were
measured in terms of factors instead of single empirical variables.

4. Accepting an important dichotomy of psychological research as
process-centered (e.g., perception, learning) vs. organism-centered (e.g.,
personality), our model and method have so far applied largely to the
latter, but untouched applications exist in the former. The applications
would consist principally, (a) of using profile similarity statistics (e.g.,
[17]; see under The Conceptual Status and Interpretation of Factors
above) to identify "types" of unitary process, i.e., to recognize the
independent processes to be studied in a taxonomy of process, and (b]
of using P technique (with lead-and-lag correlations), incremental R
technique, or even simple R technique at different process stages (as in
the learning studies of Fleischman) to throw light on the phases and
developmental patterns of psychological processes. In short there are
structural and taxonomic methods and concepts implicit in multivariate
experimental design which transcend the present local psychological
theories and have wide, permanent, and "philosophical" applicability.

The important editorial question ill}: in effect, To What Degree
Has Research in the Area Taken on a Programmatic Form? has been
answered at many points as we proceeded. It will be seen that in the

314 RAYMOND B. GATTELL

central area a very* high degree of planful programming has been under-
taken and largely realized. The Thurstones3 work in abilities had the
best qualities of research "vision" and of steady persistence in step after
step necessary to link findings in a coherent body of theory and checked
fact. The work of Guilford and his associates, on creative ability, has
had this same monumental quality, though not so far advanced; where-
as Eysenck and the London group have followed a visible, broad
pattern, even if at times moving too fast to check, consolidate, and im-
prove factor techniques per se. Again, the present writer's laboratory
started over twelve years ago a program of simultaneous factoring in
the three media of personality observation — behavior in situ (rating),
questionnaire, and objective, situational tests. This involved cross-media
factoring, developmental studies factoring at four different age levels,
checking of functional unities found in R technique by P technique,
determining of obliquities accurately enough to explore second-order
factor relations, and improving statistical significance tests for the various
modes of checking hypotheses used in these new realms.

All these aspects of the program have been brought to fruition in
some degree of empirical research. Yet among the tragedies of current
research organization we find, first, that foundations have not realized
the administrative demands, and demands on computing resources, neces-
sary if multivariate research programs are to reach effectiveness, and
secondly, that the centers in which such research proceeds can be
counted (in the world) on the fingers of one hand. Hence the scientific
peer groups necessary to check, criticize, and disseminate knowledge
among graduate students lag far behind the number desirable for
development of a healthy convergence and articulation in a scientific
system. In spite of this, and to a degree unrealized in a depressing
number of university teaching departments, there has been achieved a
considerable fraction of the programmatic research necessary for ex-
amining the integration of these -concepts and for articulating different
empirical domains.

The final editorial category for systematic presentation — a Look to
the Future in Terms of Long-range Strategy of Development — has also
been considered at each stage of our exposition, but some quite specific
summaries of more urgent research needs may now conclude this essay.
In the first place, in a strategy which envisages advance through work
in far more departments than happen to be equipped with electronic
computers and advanced statistical arts, it is necessary to supply good
factor measures, that can confidently be employed anywhere in uni-
variate research, involving factors but not factor analysis. At present,
the O-A batteries [28] are the best available stopgap, but there is crying
need for factor confirmation and intensification, i.e., for development

Personality Theory from Quantitative Research 315

of more factor saturated tests, as Sells proposes, for all eighteen general
personality factors (in objective tests), and as carried out for one
factor, recently, by Scheier (the anxiety factor, U.I. 24). Although the
present writer, from his immersion in the field of objective personality
test development, would judge that there is evidence, beyond that
summarizable in statistical P values, for confidence in the 18 dimensions
stated, many psychologists adopt a scepticism ultimately based on a
belief that nothing so tenuous as a mathematical factor can be psy-
chologically real It would be advantageous to psychology if such
skeptics would go explicitly on record with alternative statements, ex-
pressing their beliefs on the form of the psychological patterns concerned,
and set to work on checking studies. For much valuable manipulative
univariate experimental work could proceed with the advent of more
saturated, if not more definite, measurement of these factors. Unfortu-
nately ratings and, to a lesser degree, questionnaire factors are less widely
experimentally applicable and their relations to the objective test factors,
in half the cases, remain to be discovered.

One of the first needed experimental clarifications is a sorting of
factors into those of environmental and those of genetic origin. As
argued in more detail elsewhere [28], a great saving of research effort
would result if, initially, research were directed to this end as soon
as factors are validly measurable. For there is no point in entering upon
any explanation of a pattern in terms of rival learning theories if no
learning theory is going to account for the pattern. The recent genetic
studies of Beloff, Blewett, Cattell, Eysenck, Prell, and Stice [see sum-
mary in 28] already indicate that most of the variance in intelligence,
schizothymia, and Parmia vs. Threctia is hereditary, whereas most of
that in Surgency vs. Desurgency, Premsia vs. Harria, and superego
strength is environmental.

Secondly, more extensive and intensive studies are needed of the
natural life course of factors. This involves both factorings at different
cross-sectional levels to determine changes of pattern with age, as in the
work of Coan (see [28]), Gruen [37, 38], Hofstaetter [64], Peterson
(see [28]), Thurstone [90, 92], Woodrow [101]— and theoretically in
Ferguson [52] and Ahmavaara [1] — which has shown both continuity
and significant trend, and simple measures on the age trends of the single
scores for personality factors, such as have hitherto been made only for in-
telligence. The initial results already show them to have very distinctive
life courses [28]. For example, in our culture, Surgency declines steeply
between twenty and thirty years, whereas ego strength rises steadily
through most of the life course. Both these developments would help
check on hypotheses about the factors per se and also help enrich the
conceptions about factor structures in general.

316 RAYMOND B. GATTELL

Perhaps the greatest discrepancy between the pioneer promise of the
system and the necessary massive research reinforcement lies in the area
of dynamic structuring and the calculus of conflict. The possibilities for
putting most existing kinds of clinical hypothesis to a quantitative test
(after rephrasing in accordance with the more objective structural in-
dications) are obvious and manifold — and neglected. The implications
for motivational research, however, are very relevant also in the ex-
perimental and physiological study of human and animal motive, which
could benefit from more reliable measurement of drive strengths, per-
mitting the emergence of laws of learning and conflict, at present
drowned in the huge error variance (and lost also in the conceptual
vagueness of experimental measurement, where it concerns motivation
strength).

No intelligent view of current research strategy can overlook the
necessity for enlisting the help of mathematical statisticians in solving
and anticipating the problems which arise in this field as the use of
the model and the exactitude of its experimental testing increase. "An-
ticipating" is a superfluous luxury at the moment, for there is still
a backlog of problems either unsolved or insufficiently solved. In par-
ticular it is necessary to discover a statistical test for completeness of
factor extraction, to make an improvement on Bargmann's test [7]
of significance of simple structure, to solve the parallel proportional
profiles equations for factor rotation for the oblique case [29], to de-
velop a parametric test, beyond the Cattell-Baggaley s index [28], for
deciding when one factor study confirms another, and to improve on
the Oblimax analytical rotation to simple structure, now programmed
for computers, owing to the excellent work of Saunders and Dickrnan,
but still imperfect. In this part of the program the mathematician must
be persuaded to help the psychologist to psychological goals, not to
impose rigid but inappropriate mathematical perfection, as was done
with orthogonal axes. A word on models below will enlarge on this.

A realistic summary of research strategy must include graduate
teaching. It is an unfortunate fact, seldom commented upon, that the
demand in various large-scale government and industrial concerns for
Ph.D.'s competent in multivariate methods has been so great that the
normal supply of personnel to university teaching posts has been cut
off. A vicious circle is thus set up in universities in which the area is
insufficiently taught and a still more insufficient supply of teachers is
generated to teach it.

Although the multivariate analytical approach, after half a century
of rapid growth, is at the point of presenting psychologists with the first
definite set of quantitative findings on personality and motivation
structure, based on a rich harvest of concepts which are susceptible to

Personality Theory from Quantitative Research 317

accurate experimental checks, the checks are being made all too slowly
and inadequately. The possible supersession of rough clinical notions
by replicable experimental evidence on dynamic structure could gen-
erate a new phase in clinical psychology and make possible a wide
surge of manipulative experiment on the genetic, physiological, and
learning laws governing causes and consequences of this personality
structure — experiment which could not before be profitably undertaken.
The incompleteness of the wTork on factor analytic structuring itself
thus continues to present a "bottleneck" in the flow of objective per-
sonality research.

Doubtless as more universities recognize the basic logical connection
of multivariate and clinical research, clinical resources will help to solve
the present research supply problem by training a new type of re-
searcher in clinical and personality areas, simultaneously competent in
clinical observation and multivariate experimental designs and calcula-
tions. Meanwhile few such graduate students are produced, and fewer
surmount the temptation of the market place, so that such adequately
trained persons constitute perhaps not one in ten of the people struggling
with the complex problems of clinical and personality research. The
clinical practice area will continue to be the largest source of good data
for personality research. Consequently, any practical forecast of how
soon one may expect the predicted broad development of new laws
based on structured personality measurement, rising beyond the level
of the present clinical, and general, nonquantitative theory (based on
perception by the naked eye alone), depends on two matters outside
science per se, and difficult to estimate, namely: (1) the amount of lag
in teaching departments in switching to new training goals for clinicians,
counsellors, and other major sources of researchers in personality, and
(2) the growth of the social and economic organization of a more
complex type of research institution, involving coordination of clinics,
electronic computers, centers for basic personality research, machinery
for assembling larger samples of subjects (from more comparable and
controlled populations), research committees to appraise factor match-
ing and factor batteries, etc. This larger organization becomes a neces-
sity for effective work in multivariate research (though never really
needed in univariate work) because there is a far greater tendency for
any apparently local problem in multivariate research actually to in-
volve the whole area, i.e., to involve simultaneously many variables and
many types of subject.

To the editorial request that each contributor conclude by extend-
ing reference to "Barriers Blocking General Theoretical Advance in Psy-
chology" the present writer would add, in addition to the difficulties
in the socioeconomic organization of research and in graduate in-

318

RAYMOND B. CATTELL

straction, the psychologist's own preoccupation with theory. It needs
no very piercing clinical eye to see that psychology suffers from a sense
of inferiority — as it compares its theoretical development with that of
the physical sciences. A fair number of psychologists have reacted to
this by importing pretentious theories from outside, though they have
no natural organic relation to psychological data. Some grotesque
productions have ensued; these merely distract psychology students who
are interested in theory (not too numerous, incidentally) from the theory
that is growing more modestly out of local laws and newly observed,
peculiar regularities in quantitative psychological research data.

A slight regression of psychology toward reattachment to its step-
mother, philosophy, is evident in the recent profusion of mathematical
models. Like any other machine, the model must be the servant of the
psychologist, not his master. Speculation on models far beyond any
psychological exemplification or means of checking by data is no service
to psychological theory. The playground for indulging mathematical
models is infinite, but as with trial and error in biological mutations, the
percentage proving adapted to reality will be very small. Consequently,
the theorist should exercise some choice from the start, in terms of
adopting models suggested by psychological data, rather than those
appealing to the mathematician. The oblique factor analytic model,
central in the present psychological theory', would quickly have died of
neglect if left to most mathematical statisticians, for it is beset with
unpleasantly unsolvable issues, with assumptions which are often not
exactly met, and with some properties as irrational as *. But the psy-
chologist has developed it because he is an observer and a scientist first,
and only secondarily a mathematical theorist.

APPENDIX: THE CONCEPTS OF VARIABLE DENSITY AND FACTOR
ORDER

For most of the more specialized, complex theoretical or statistical
developments touching the above outline of the field the reader can
be, and has been, referred to publications elsewhere. Since at the mo-
ment of writing no source is available to deal with one vital specialized
issue — variable sampling — it seems best to handle that here. The no-
tions of a total available population of variables, and of a means of
taking a stratified sample of it, underlie the concept of the personality
sphere discussed earlier and are necessary to a true rationale for dis-
tinguishing first- and higher-order factors [28].

It is an assumption of factor analysis that no factor affects all
variables of behavior and that boundaries can, therefore, be drawn in
some conceptualized "behavioral space" (or area of variables) delimit-

Personality Theory from Quantitative Research 319

Ing the realm of action of any factor. This assumption is primarily
involved in simple structure, but it is also involved, for example, in
Guttmann's "radex" theory of factor structure, and it also certainly
seems to fit any empirically obtained factors, by practically any system
of rotation.

First-order factors necessarily affect larger areas than specific factors
and second-order factors than first-order. Thus in Fig. 3 the interrapted-

FIG. 3. Emergence of factors in relation to choice of variables. After
Cattell.

Hne squares 'can be considered to represent specific factors, a, b3 cy dy e,
the continuous-line areas the first-order factors, A, B, and C, and the
heavy continuous the second-order factors, Alpha and Beta.

Now it will be seen that if we factor variables 1, 2, 4, 6, 8, and 9,
we shall first obtain primary factors A, B3 and C (B and C both loading
on 8) and specifics a} b, c, d, e, and /. On factoring A and B and C
we shall obtain (providing there are enough other primaries to com-
plete the definition) two second-order factors, Alpha and Beta. If, on
the other hand, we had happened to begin with variables 1, 8, and 9
(plus such others analogously placed as are necessary to define two

320 RAYMOND B. CATTELL

factors), we should have reached factors Alpha and Beta directly,
without realizing they are second order. Conversely, if it were possible,
as it often is, to get two variables so similar as in the pairs 2 and 3, 4 and
5, 6 and 7? and 9 and 10, without actually being identical, our factor-
ing would have given the specifics b, c, d, and / as first-order common
factors.

Now, in the absence of any workable concept of "density of variable
sampling,55 it certainly happens in research that we sometimes change
scale in this way without knowing it. On the other hand, it is unlikely
that we do frequently, and the most probable state of affairs is that we
mix first- and second-order factors for primary and specific, but pseudo-
common) in a single research. For example, if one started a factoriza-
tion with variables 1, 2, 3, 7, and 9 in the above diagram, factor A
would be found by its loading of 1, 2, and 3. But unless 1, 2, or 3
contained some variance in B and C (which by definition they do not)
factors B and C would not appear, since there is only one representative
of each. On the other hand, the second-order factors Alpha and Beta
would appear, to the confusion of matching efforts in the primary
realm.

Much of the difficult}7 and disagreement in present research findings
is probably owing to this source — quite apart from the related effect, as
pointed out by Ahmavaara [11, that primary factors found at lowTer
population ages may "fine out" into two or more distinct factors at
higher ages. Careful records of factors found with different sets of
variables, finally put together with every possible cross comparison,
might do something to solve this difficulty. But no one has succeeded in
so unraveling the tangle, and it is more likely that we shall succeed
better by getting an independent concept of "density of representation"
of variables, fixing this at a definite figure for all researches to be inte-
grated. This concept implies the notions of both (a] a total population
or area of possible variables and (b) the distance apart of any twro
variables.

Now either the personality sphere [14, 231 or the time sampling of
human behavior [28] can give us the former, but where do we get the
latter? The familiar notion of the distance apart of two variables in
factor analytic space will not help, for this is a dependent value, and we
have to compare this with the new, independent concept of distance, to
check on whether our factor analysis is right. Further, the notion of a
total area of human behavior is almost certainlv going to depend on the
assumption that items of behavior gathered according to a certain
operational procedure are equidistant from one another. If so this vital
difference will exist between the total sampled variable space and the

Personality Theory from Quantitative Research 321

resultant factor analytic space: that whereas the variables are spaced
with equal density throughout the former, they are very definitely con-
centrated largely in hyperplanes in the latter.

The search for a concept of degree of similarity (or distance) of
variables must thus proceed outside, and independently of, the factor
analytic model. The possible alternatives are a functional or a phenorn-
enological basis. By the former is meant any system of classifying
situations and responses which depends on the functional relations of
these situations to man, i.e., on their relations to human functioning.
By the latter is meant a classification supposedly depending on the real
properties of the objects, detached as far as possible from human
perception.

A number of functional classifications of variables according to
similarity, which do not directly depend on correlational and factor
analytic measures, can be suggested. For example, the likeness of
situations might be measured in terms of the amount of learning
transfer from one to another. Another possibility would be in terms of a
threshold of discrimination in perception of these situations. For ex-
ample, if as far as a person can see, two tests are really demanding the
same thing of him, then we should not consider them one jnd apart. Or
again, we might take the degree to which fatigue transfers from one
type of reaction to another. Or yet again, we might take the frequency
with which interest in one activity can be substituted stably for interest
in the other.

There are undiscussed difficulties in most of these, such as the fact
that the variables are not defined by stimulus-response but require also
reference to the mode of scoring. But a more general objection is that
any functional classification might prove to be in some degree related
to any other, and therefore, to the factor analytic one. On the other
hand, the phenomenological basis is open from the beginning to the
criticism that human reactions to situations depend less on the real
nature of the situation than on the personal or cultural history of condi-
tioning to them. Consequently, the similarity of situations in any con-
ceivable psychological sense is likely to have little relation to any pos-
sible index of phenomenological similarity. Thus, the similarity of
variables can at present be envisaged only on a functionalistic basis.
We should not abandon this avenue, despite the suspicion of its nonin-
dependence of factor analysis, until the relation has been empirically
examined. But this will take time, for though we have ample data on
correlational closeness of variables we have virtually none on transfer
of learning or fatigue, perceptual similarity, or motivational equivalence
in relation to variables and factors.

322 RAYMOND B. CATTELL

REFERENCES

1. Ahmavaara, Y. Transformation analysis of factorial data. Ann. Acad.
ScL Gen., 1954. 88 (2 i , 1-150.

2. Allport, G. W., & Odbert, H. S. Trait-names: a psycholexical study.
Psychol. Monogr., 1936, 47, 211 ? 171.

3. Anderson. E. E. Inter-relationships of drives In the male albino rat:
I. Intercorrelatlons of measures of drives. /. comp. Psychol., 1937, 24, 73-118.

4. Anderson. H. H., & Anderson, G. H. An introduction to projective
techniques. Englewood Cliffs, N.J.: Prentice-Hall, 1951.

5* Andrews, T. G. A factorial analysis of responses to the comic as a
study In personality. /. gen. Psychol., 1943, 28, 209-226.

6. Bargmann, R. Eine Modelstudie zur Erlauterung der Wirksamkeit
Faktoren-analytischer Methoden, Monogr. Hochschule Int. Padag. Forsch.,
Frankfurt a. m., June, 1955. Pp. 1-34.

7. Bargmann, R. Signlfikanzuntersuchungen der Einfachen Struktur In
der Faktoren Analyse. Mitteilungsblatt f. Math. Statist. Wurzburg: Sonder-
druck, Physica-Verlag, 1955. Pp. 1-24.

8. Bell, R. O. A comparison of profile prediction methods and multiple
correlation. Unpublished doctoral dissertation, Stanford Univer., 1951.

9. Brogden, H. E. A factor analysis of forty character tests. Psychol.
Monogr., 1940, 234, 35-55.

10. Burt, C. The factorial analysis of emotional traits. Charact. & Pers.,
1939, 7, 238-254, 285-299.

11. Cattell, R. B. Temperament tests: I. Temperament. Brit. J. Psychol.,
1932, 23, 308-329.

12. Cattell, R. B. "Parallel proportional profiles" and other principles
for determining the choice of factors by rotation. Psychometrika, 1944, 9,
267-283.

13. Cattell, R. B. The diagnosis and classification of neurotic states:
a reinterpretation of Eysenck's factors. /. nerv. ment. Dis., 1945, 102, 576-
589.

14. Cattell, R. B. The description and measurement of personality. New
York: World Book, 1946.

15. Cattell, R. B. Personality factors in response to humor. /. abnorm.
soc. Psychol, 1947, 42, 402-421.

16. Cattell, R. B. The ergic theory of attitude and sentiment measure-
ment Educ. Psychol. Measmt, 1947, 7, 221-246.

17. Cattell, R. B. rp and other co-efficients of pattern similarity. Psy-
chometrika, 1949, 14, 279-298.

18. Cattell, R. B. Principles of construction of projective or "Ego-de-
fense53 tests of personality. In H. H. Anderson, Projective techniques. Engle-
wood Cliffs, N.J.: Prentice-Hall, 1950. Chap. 2.

19. Cattell, R. B. The main personality factors in questionnaire self-
estimate material /. soc. Psychol., 1950, 31, 3-38.

20. Cattell, R. B. Personality: a systematic theoretical and factual study.
New York: McGraw-Hill, 1950.

Personality Theory from Quantitative Research 323

21. Cattell, R. B. The discovery of ergic structure in man in terms of
common attitudes. /. abnorm. Psychol., 1950, 45, 598-618.

22. Cattell, R. B. The three basic factor analytic research designs —
their interrelations and derivatives. Psych. Bull, 1952, 49 (5), 499-520.

23. Cattell, R. B. Factor analysis: an introduction and manual for the
psychologist and social scientist. New York: Harper, 1952.

24. Cattell, R. B. A guide to mental testing. (3d ed.) 1602 Goronado
Dr., Champaign, 111.: IPAT, 1953.

25. Cattell, R. B. Personality structures as learning and motivation
patterns: a theme for the integration of methodologies. In Kentucky Sym-
posium on Learning theory, personality theory and clinical research. New
York: Wiley, 1954. Chap. 7.

26. Cattell, R. B. The principal replicated factors discovered in objective
personality tests. /. abnorm. soc. PsychoL, 1955, 50 (3), 291-314.

27. Cattell, R. B. The chief invariant psychological and psycho-physical
functional unities found by P-technique. /. din. Psychol., 1955, 11, 319-343.

28. Cattell, R. B. Personality and motivation structure and measure-
ment. New York: World Book, 1957.

29. Cattell, R. B. Factor rotation for proportional profiles: analytical
solution and an example. Brit. J. stat. Psychol, 1955, 8, 83-92.

30. Cattell, R. B., & Anderson, J. C. The measurement of personality
and behavior disorders by the I PAT Music Preference Test. /. appl.
Psychol, 1953, 37 (6), 446-454.

31. Cattell, R. B., & Baggaley, A. R. The salient variable similarity index
— s — for factor matching. Univer. of 111., Laboratory of Personality As-
sessment and Group Behavior, Adv. Publ. No. 4, 1954.

32. Cattell, R. B., & Baggaley, A. R. The objective measurement of
attitude motivation: development and evaluation of principles and devices.
/. Pers.y 1956, 24, 401-423.

33. Cattell, R. B., & Baggaley, A. R. A confirmation of ergic and
metanergic structure in attitudes objectively measured. Austral. /. PsychoL,
in press, 1958.

34. Cattell, R. B., Cattell, A. K. S., & Rhymer, R. M. P-technique
demonstrated in determining psycho-physiological source traits in a normal
individual. Psychometrika, 1947, 12, 267-288.

35. Cattell, R. B., & Cross, P. Comparison of the ergic and self-sentiment
structures found in dynamic traits by R- and P- techniques. /. Pers.} 19523
21 (2), 250-270.

36. Cattell, R. B., Dubin, S. S., & Saunders, D. R. Verification of hy-
pothesized factors in one hundred and fifteen objective personality test de-
signs. Psychometrika, 1954, 19, 209-230.

37. Cattell, R. B., & Gruen, W. Primary personality factors in eleven
year old children by objective tests. /. Pers., 1955, 23 (4), 460-478.

38. Cattell, R. B., & Gruen, W. Primary personality factors in the
questionnaire medium for children eleven to fourteen years old. Educ.
psychol Measmts 1954, 14, 50-76.

324 RAYMOND B. CATTELL

39. Cattell, R. B., & Saunders, D. R. Musical preferences and personality
diagnosis: I. A factorization of one hundred and twenty themes. /. soc.
PsychoL, 1954, 39, 3-24.

40. Cattell, R. B., & Saunders, D. R. Beitrage zur Faktoren-Analyse der
Personlichkeit Z. exp. angew. Psychol, 1955, 2 (3)? 325-357.

41. CattelL R. B.? & Stice, G. L. "The psychodynamics of small groups,"
final report on research project XR1 72-369, Contract ZN80nr-79600?
Human Relations Branch, Office of Naval Research, 1953. Pp. 1-207.

42. Cattell. R. B., & Wenig, P. Dynamic and cognitive factors con-
trolling misperception. /. abnorrn. soc. PsychoL, 1954, 47 f 4) , 797-809.

43. Coombs, C. H., & Satter, G. A. A factorial approach to job families.
Psychometrika, 1949, 14, 33-42.

44. Cronbach, L. J., & Gleser, G. C. Assessing similarity between profiles.
Psychol. Bull, 1953, 50, 456-473.

45. Crutcher, R. An experimental study of persistence. /. appl. PsychoL,
1934, 18, 406-417.

46. Cureton, E. E. A note on the use of Burt's formula for estimating
factor significance. Brit. J. PsychoL, 1955, 8, 28.

47. Das, Rhea. An investigation of attitude structure and some hypoth-
esized personality correlates. Unpublished doctoral dissertation, Univer. of
Illinois, 1955.

48. Ekman, G. On topological and dimensional systems of reference in
describing personality. Ada Psychol., 1951, 8, 1-24.

49. Eysenck, H. J. The scientific study of personality. London : Routledge
ftKegau Paul, 1952. Pp. 26, 61, 148, 161, 162, 206, 299.

50. Eysenck, H. J. The structure of humon personality. London:
Methuen, 1953.

51. Eysenck, H. J., & Prell, D. B. The inheritance of neuroticism: an
experimental study. /. ment. ScL, 1951, 97, 441-465.

52. Ferguson^ G. A. On learning and human ability. Canad. J. Psychol.,
1954,8(2).

53. Fiedler, F. "The influence of leader-keyman relations on combat
crew effectiveness,53 Tech. Report No. 9, Univer. of Illinois, Group Effective-
ness Laboratory, 1954.

54. Fiske, D. W. Consistency of the factorial structures of personality
ratings from different sources. /. abnorm. soc. Psychol., 1949, 44, 329-344.

55. Flanagan, J. C. Statistics for the clinician, /. din. PsychoL, 1950, 6,
74-76.

56. French, J. W. The description of personality measurements in terms
of rotated factors. Princeton, N.J.: Educ. Test. Serv., 1953.

57. French, J. W. Selected tests for reference factors. Princeton, N.J.:
Educ. Test. Serv., 1953.

58. Fruchter, B. Introduction to factor analysis. Princeton, N.J.: Van
Nostrand, 1954.

59. Guilford, J. P., & Martin, D. R. The Guilford-Martin Personality
Questionnaire. Beverly Hills, Calif.: Sheridan Press, 1950.

Personality Theory from Quantitative Research 325

60. Gullford, J. P., Schnelderman, E., & Zimmerman, W. S. The
Guilford-Schneiderman-Zimmerman Interest Survey. Beverly Hills, Calif.:
Sheridan Press, 1948.

61. Harlow, H. Motivational forces underlying learning. In Kentucky
Symposium on Learning theory, personality theory and clinical research.
New York: Wiley, 1954. Pp. 36-53.

62. Hartshorne, H., May, M. A.3 & Shuttleworfh, F. K. Studies in the
organization of character. New York: Macmillan, 1930.

63. Haverland, E. M. The application of an analytical solution for
proportional profiles rotations to a box problem and to the drive structure
in rats. Doctoral dissertation, Univer. of Illinois, 1954.

64. Hofstaetter, P. R. The changing composition of intelligence: a study
in T-technique. /. genet. Psychol., 1954, 85, 159, 164.

65. Horst, P. The prediction of personal adjustment. Soc. Sci. Res.
Council Bull., 1941, No. 48.

66. Hull, C. L. Variability in amount of different traits possessed by
the individual. /. educ. Psychol, 1927, 18, 97-106.

67. Kelley, E. L., & Fiske, D. W. The prediction of performance in
clinical psychology. Ann Arbor, Mich.: Univer. Michigan Press, 1951. Pp.
80, 118.

68. Lee, M. C. Configural vs. linear prediction of collegiate academic
performance. Doctoral dissertation, Univer. of Illinois, 1956.

69. Lubin, A. "A methodological study of configural scoring," Tech.
Notes No. 42, A.G.O., Personnel Research Branch, 1954.

70. Lubin, A., & Osburn, H. G. The general theory of pattern analysis
for predicting quantitative criteria. San Francisco: APA Division 5, Sept.,
1955.

71. McClelland, D. G., et al. The achievement motive. New York:
Appleton-Century-Crofts, 1953.

72. McNemar, Q. Practice and "general" motor ability. /. gen. Psychol.,
1930, 3, 67-97.

73. McQuitty, L. L. A statistical method for studying personality
integration. In O. H. Mowrer (Ed.), Psychotherapy: theory and research.
New York: Ronald, 1953. Chap. 16, pp. 414-462.

74. McQuitty, L. L. Pattern analysis illustrated in classifying patients
and normals. Educ. psychol. Measmnt,, in press, 1956.

75. McQuorquodale, K., & Meehl, P. E. On a distinction between
hypothetical constructs and intervening variables. Psychol. Rev., 1948, 55,
95-107.

76. Meehl, P. E. Clinical versus statistical prediction. Minneapolis,
Minn.: Univer. Minn. Press, 1954.

77. Mowrer, O. H. Neurosis: a disorder of conditioning or problem
solving? In E. J. Kempf (Ed.), Comparative conditioned neuroses. New
York: N.Y. Acad. Sci., 1953.

78. Murray, H. Explorations in personality. New York : Oxford Univer.
Press, 1938.

326 RAYMOND B. CATTELL

79. Neuhaus, J. G., & Wrigley, C. The quartimax method. Brit. /. stat.
PsychoL, 1954, 7, 81-91.

80. Osgood, G. E., & Suci, G. J. A measure of relation determined both
by mean difference and profile information. PsychoL Bull, 1952, 49, 251-
262.

81. Rao, C. R. Advanced statistical methods in biometrics research. New
York: Wiley, 1952.

82. Rethlingshafer, D. The relation of tests of persistence to other
measures of continuance of action. /. abnorm. soc. PsyckoL, 1942, 37, 71—
82.

83. Ryans, D. G. The meaning of persistence. /. gen. PsychoL, 1938, 19,
79-96.

84. Saunders, D. R. Item analysis of interaction variance to produce
a criterion key for item pairs. Res. Bull.., No. 55-28. Princeton, N.J.: Educ.
Test Serv., 1955.

85. Sears, R. R. Survey of objective studies of psychoanalytic concepts.
Bull Set. Res. Council, 1943, No. 51.

86. Sears, R. R. Experimental studies of projection 23 ideas of reference.
/. soc. PsychoL, 1937, 8, 389-401.

87. Spence, K. W. The nature of theory construction in contemporary
psychology. PsychoL Rev., 1944, 57, 47-68.

88. Symonds, P. M. Diagnosing personality and conduct. New York:
Appleton-Century-Crofts, 1931.

89. Thornton, G. R. A factor analysis of tests designed to measure
persistence. PsychoL Monogr., 1939, 51 (3), 1-42.

90. Thurstone, L. L. Multiple factor analyses. (3rd ed.) Chicago:
Univer. Chicago Press, 1949.

91. Thurstone, L. L. Primary mental abilities. Chicago: Univer. Chicago
Press, 1949. P. 282.

92. Thurstone, L. L. A factorial study of perception. Chicago: Univer.
Chicago Press, 1944. P. 282.

93. Torr, D. V. A factor analysis of 49 interest variables. USAF Bull.
Human Relat. Res. Cent., 1943, No. 53-67.

94. Tucker, L. R. Relative predictive efficiency of multiple regression
and unique pattern techniques. Microfilm Abstr., 1951, 11 (2), 439-440.

95. Tucker, L. R. The objective definition of simple structure in linear
factor analysis. Psychometrika, 1955, 20, 209-255.

96. Vemon, P. E. The structure of human abilities. London: Methuen,
1950. Pp. 113,295.

97. Ward, J. EL "An application of linear and curvilinear joint func-
tional regression in psychological prediction," Res. Bull. AFPTRC-TR-54-
86, AFPTRC, Lackland AFB, Lackland, Texas, 1954.

98. Wenger, M. A. Studies of automatic balance in Army Air Force
personnel. Comp. psychol Monogr., 1948, 19, 1-111, 195.

99. Wenig, P. Relative roles of naive, autistic, cognitive and press com-
patibility misperception and ego defense operations in misperception. Un-
published master's thesis, Univer. of Illinois, 1952.

Personality Theory from Quantitative Research 327

100. Williams, J. R. A comparison of P- technique measures and clinical
evaluations of adjustment and conflict. Unpublished doctoral dissertation,
Univer. of Illinois, 1958.

101. Wittenborn, J. R.3 Madler, G., & Waterhouse, I. K. Symptom
patterns in youthful mental hospital patients. /. din. PsychoL, 1951, 53,
323-327.

102. Woodrow, H. Inter-relationships of measures of learning. /.
PsychoL, 1940, 10, 49-73.

PSYCHOGENETIC STUDIES OF TWINS

- FRANZ J. KALLMANN
New York State Psychiatric Institute, Columbia University

The Place of Psychogenetics In Science 328

General Methodological Principles 330

Determination of Zygosity 332

Ascertainment and Analysis of Twin Samples 334

Procedural Limitations and Advantages 338

Serial Twin Data on Intellectual and Personality Variations 343

Serial Twin Data on Psychopathological Variations 350

Prospects for Future Twin Research 354

Schematic Recapitulation 356

General methodological principles 356

Determination of zygosity 356

Ascertainment and analysis of twin samples 356

Procedural limitations and advantages 357

Serial twin data on intellectual and personality variations 357

Serial twin data on psychopathological variations 357

Prospects for future twin research 357

References 357

THE PLACE OF PSYCHOGENETICS IN SCIENCE1

Psychology and the biological discipline of human genetics meet
on the common ground of a concern with the variable dynamics of human
behavior as exhibited in individuals or groups. The mutual interests
of the two sciences lie in the very specific characteristics that cause men
to strive and create, to maintain health or succumb to adversity, to
choose a proper mate, to work, reproduce, and grow old, to die in harness
or in the feeble shadows of retirement [48]. There are biological founda-
tions for each of these functions, and all of them are genetically con-
trolled.

Although a major objective of psychological research into personality

1 The assistance of Dr. Arnold Kaplan in preparing the bibliographical ma-
terial of this report is gratefully acknowledged.

328

Psycho genetic Studies of Twins 329

is to appraise quantitative and qualitative differences in traits and at-
titudes from an "ethnocentric" standpoint [23], the aim of genetic studies
is to search for the basic causes. The changing behavior patterns and
adjustment problems of individuals and societies, and the various modes
of adaptability to a multitude of technical advances of modern man
depend on determining basic factors, which, in turn, follow certain
laws.

Some of the underlying principles are elementary, others highly
complex. A few are still the subject of much hypothetical controversy,
if only because "the cleavage between natural and social science ... is
a cleavage between substance and action, body and soul, the objective
and the subjective" [34]. Though the science of human genetics is less
than sixty years old, and the twin-study method is the only quasi-ex-
perimental genetic procedure available in man (arranged by Nature
rather than human ingenuity), a wealth of empirical data has been
produced for man's study of himself and his origins [12].

Against this background of "the collective properties that describe
the living" [34] and of "the myriad elements constituting the life cycle
of a human organism" [1], an attempt will be made here to review those
comparative results, methodological principles, and conceptual implica-
tions of twin studies which fall into the broad sector of psycho genetics
(physiological, psychological, and psychiatric genetics). The research
method to be described and critically evaluated has led to data having
fundamental import for any theory of human behavior. Therefore,
though the concern here differs from that of most contributions to the
present study, the method under analysis has distinct significance for an
assessment of the systematic status of contemporary psychology. Needless
to say, however, the analysis of a research method, rather than a body
of systematic statements, cannot be carried out in strict conformity to the
rubrics of the discussion scheme suggested for these volumes.

It is worth emphasizing that the need to depart from the discussion
outline in certain ways is no reflection on the relevance of psychogenetic
twin data to the human sciences. The character and problematic ob-
jectives of the method used for generating such data are dictated by
general questions basic to the human sciences. Although it is true that
twin studies represent a highly specialized research method, which has
born fruit in the garden of human genetics, the facts garnered there call
for certain generalizations which must condition the content of any psy-
chological theory of fundamental intent. In the long run, no theory of
human behavior or personality can evade the detailed implications of
psychogenetic knowledge with regard to such matters as basic problem
definition, selection of variables and general causal £<model," treatment
of basic behavior "processes" (e.g., perception, learning, motivation),

330 FRANZ J. KALLMANN

and of development, individual and species differences, and many other
issues. Obvious as these facts are, perhaps it is fair to recall that not too
long ago it was not uncommon for psychological theorists to operate
in something close to a genetic vacuum, and even today one sees
occasional evidence of theorists attempting to settle the specifically
genetic issues confronting them by fiat rather than reference to available
knowledge.

Necessarily, refinement of the usefulness of twin studies in exploring
the complexities of human behavior paralleled the gradual advance of
human genetics toward a status of scientific respectability. This inter-
dependent development sparked a growing awareness in all behavioral
sciences of the excellent opportunities afforded by twins, revitalizing in
turn the interest in the predominantly genetic aspects of those sciences.
At the same time, however, it introduced into psychogenetics a great
many controversial issues and procedural problems peculiar to disciplines
beclouded by conflicting ideologies. Arising from the widespread con-
viction that scientific thinking ought to conform to political thinking,
the tension in the atmosphere was Increased by the notion that there is
some basic conflict between religious tenets and the scientific principles
of human genetics.

It will always be to the credit of the small and widely scattered
phalanxes of twin researchers in many countries that they made a con-
scientious and sustained effort toward establishing psychogenetics as an
Ideologically unshackled discipline within the behavioral sciences.

GENERAL METHODOLOGICAL PRINCIPLES

Historically, it is of interest that long before medicine developed
into a full-fledged science, artists and scholars focused attention on twin
births and attempted to find an explanation for this phenomenon [33].
Ancient mythology contains many references to twin divinities. The
Babylonians and Assyrians introduced twins in astronomy, thus giving
rise to innumerable legends and horoscopes. Hippocrates believed twins
were conceived by the division of the sperm into two parts, with each
part penetrating one of the two uterine horns. Cicero commented on
what Diogenes had to say about twins and the astrologers who, then as
now, insisted that temperament is determined by the influence of the
stars. Aristotle and Empedocles expressed the idea that double mon-
strosities might originate from a phenomenon of codevelopment (partial
fusion), and Galen thought that excess heat in the uterus might split the
sperm, thus originating two or more formations. During the many
centuries dominated by the Arabic and Salernitan schools of medicine,
"writers merely reiterated the classic ideas about twins" [33].

Psycho genetic Studies of Twins 331

Although Viardel observed in 1671 that uniovular twins were always
of the same sex, it was not until the second half of the nineteenth century
that the phenomenon of twinning ceased to be regarded as a biological
whim which aroused either a sense of alarm or idle curiosity. It was
Sir Francis Galton who had the foresight to recognize the usefulness of
the lives of twins as a research tool in the service of science. His two
treatises. The History of Twins as a Criterion of the Relative Powers of
Nature and Nurture (1876) and Inquiries into Human Faculty and Its
Development (1883), were classical contributions to the implements of
psychogenetics.

Since the rediscovery of Mendel's ingenious theories regarding the
"unblending" behavior of stable genetic units in organic inheritance
(1900), investigators in many countries have availed themselves with
increasing frequency and better techniques of the unique opportunities
presented by the regular occurrence of two genetically different types of
twins — those derived from one fertilized ovum, and those derived from
two fertilized ova. Whereas one-egg twins are always of the same sex,
two-egg twins may be of the same or of opposite sex.

In the original version of the twin-study method, the comparison of
observable similarities and dissimilarities in the histories of genetically
similar or dissimilar genotypes is limited to twin subjects. This procedure
requires access to a representative series of one-egg and two-egg twins,
of either or different sex, presenting evidence of a diagnostically wrell-
defined trait to which the investigative principles of the proband method
[97] can be applied.

In another version, observational or experimental data are obtained
from a few well-selected pairs of one-egg twins whose aptitudes, physio-
logical reactions, or adjustive patterns can be compared under different
life conditions or in response to different methods of planned manage-
ment. This procedure has been used by numerous investigators, especially
by Gesell and Thompson [36], and is called the co-twin-control method.

In a third version called the twin-family method [49], the collection
of comparative data is extended to complete sibships of twin index
cases and their parents. The six dissimilar sibship groups compared in
this manner are one-egg twins, two-egg twins of the same sex, two-egg
twins of opposite sex, full sibs, half-sibs, and step-sibs. This procedure
makes it possible to combine the study of twins with the investigative
principles of the census, proband, and sibling methods, as well as with
special pedigree studies, thereby affording an excellent opportunity to
investigate intrafamily variations with a minimum of uncontrolled
variables.

The advantages of such a combined procedure are most apparent in
the study of traits which present complex sampling problems and require

332 FRANZ J. KALLMANN

comparisons in both cross-sectional and longitudinal directions. In fact,
so broad is the scope of the twin-family method that it fulfills nearly all
the requirements specified by Cattell [15] for the use of the multiple
variance method in investigations of functionally, but not necessarily
genetically, unitary traits falling into the normal range of personality
development. In his scheme, measurable test data are obtained from five
different populations, hi pairs : ( 1 ) a sample population of one-egg twins
in their own families, (2) siblings in their own families, (3) siblings with
each member of the pair in a different family, (4) unrelated persons in
pairs in the same families, and (5) unrelated persons in different families.

DETERMINATION OF ZYGOSITY

For determining the zygosity of same-sex twins, the present method
of choice is a refined version of the similarity method, originally de-
veloped by two veteran twin researchers, Siemens [77] and Von
Verschuer [92], The fetal-membrane method is no longer in use, since it
is now known that not all one-egg pairs are born with only one placenta.

In applying the modem similarity method, the comparison of such
usually variable physical characteristics as facial features, dental specifica-
tions, ear lobe form, and pigmentation of hair and eyes is supplemented
by a careful analysis of fingerprints and blood group data, the most
reliable criteria for distinguishing one-egg and two-egg twins. Other
morphological traits or metric characters cannot be relied upon per se,
especially in the presence of a grossly pathological condition in one
member of a pair. If dermatoglyphic and hematological data are in-
decisive in a scientifically important case, it may be advisable to resort
to reciprocal skin grafts [70]. Full-thickness homografts are not successful
in two-egg twins, although initial takes may last three to four weeks.

In the hematological analysis, it may be borne in mind that a given
pair of twins cannot be monozygotic if the blood groups are different.
If the blood groups are the same, however, the twins may or may not
be monozygotic. Procedural accuracy requires, of course, that in same-
sex pairs found to be similar with respect to the major ABO and Rh
factors, blood typing is continued until a difference appears or until
all available antisera (M-N, S-s, Duffy, Kell, Lutheran, and so forth)
have been tried.

The main disadvantage of hematological procedures is that they are
rather expensive and depend on the availability of both twins. There are
clinically important traits which, by their very nature, take the research
subject out of the reach of laboratories.

The dermatoglyphic analysis rests upon the fact that fingerprints
conform to one of three basic types, each of which is largely determined

Psychogenetic Studies of Twins 333

by heredity: whorls, loops, and arches. Whenever possible, the analysis
should be extended to both qualitative and quantitative aspects, although
the qualitative analysis may often suffice for practical purposes. By means
of quantitative procedures, however, dermatoglyphics alone guarantee a
maximum degree of reliability.

Of the three quantifying measures used, the simplest one is the sum
of the homolateral ridge-count differences [30], a difference of more than
40 being strongly suggestive of dizygosity. In Wendt's individual pattern
score or Musterwert [98] — a measuring device of almost equal simplicity
— each pattern is classified according to certain objective criteria and
scored from one to seven. A difference of more than five in the twins'
total score is strongly indicative of dizygosity.

Of at least the same diagnostic value are the scores obtained with
Slater's discriminant function [81], the most complex test devised. In
this method, whorls provide two counts each, loops one, and arches none.
The following five characteristics are calculated from the number of
ridges intervening between the core of a whorl and its triradii to one
and the other side, or between the tip of the innermost ridge of a loop
and the triradius :

1. The difference in total count between right and left sides, both
members of the twin pair being taken together, expressed as a proportion
of the summed total counts.

2. The difference in total count between one twin and the other,
both right and left sides being taken together, as a proportion of the
summed total counts.

3. The correlation coefficient between right and left sides, pairing
digit with digit, radial counts against radial counts, ulnar against ulnar,
both twins being taken.

4. The correlation coefficient between twins, pairing digit with
digit, the right hand of one twin with the right hand of the other, and
left with left.

5. The crossed correlation coefficient between twins, proceeding as
above, but taking the right hand of one twin with the left hand of the
other and vice versa.

According to this method, a diagnosis of dizygosity is indicated by
scores over two, whereas a score below minus one is indicative of mono-
zygosity.

In recent years, a great deal of work has been done to improve the
statistical procedures, which are employed in the diagnosis of zygosity,
by means of multiallele systems such as the blood groups. The choice
of the method depends on the manner in which the genetic system is
used [87]. The most common area of application is the assessment of
dizygosity in twin samples, since same-sex twins with the same blood

334 FRANZ J. KAIXMANN

groups may be dizygotic. If the chances of two particular twin partners
being dizygotic are to be estimated individually, the amount of hema-
tologically verifiable information may be expected to vary considerably
from family to family [69].

The probability of monozygosity for concordant twins can be cal-
culated either with or without reference to the actual phenotypes which
the concordance involves [29]. When the zygosity of a particular set of
twins is under consideration, the appropriate procedure is to calculate the
probability of monozygosity with reference to the phenotypes involved.
However, when the given probability quotient is to be obtained for a
random pair of twins or when the adequacy for diagnostic purposes of a
proposed series of genetic phenotypes is the issue in question, it has
been suggested that the probability of total concordance be calculated
without reference to the actual phenotypes. The appropriate formulas
have been rearranged by Button, Clark, and Schull [87].

With current knowledge regarding the essentials of the similarity
method so remarkably improved, serious doubt is cast upon the potential
usefulness of twin data published without clearly substantiated zygosity
classifications.

ASCERTAINMENT AND ANALYSIS OF TWIN SAMPLES

The scientific value of comparative studies, which are based on an
unrepresentative or improperly analyzed series of a few sets of multi-
zygotes, is even more questionable.

It will always be possible, for instance, to find some one-egg twins
who are distinguished by discordance as to a well-known or etiologically
obscure trait of predominantly gene-specific origin. Apart from the ex-
pected occurrence of phenocopies (nonhereditary variation usually pro-
duced by a clearly defined mutant), it is worth remembering, however,
that a genetically determined trait may be neither symmetrical in its
phenotypic expression nor completely penetrant [2, 3, 21, 55]. Just as
it is erroneous to ascribe an observed lack of penetrance of a certain
gene effect to the action of environmental factors alone, so would it be
a mistake to doubt the primary randomness of developmental processes
or of nonadaptive right-left asymmetries in embryonic development,
which take place on a biological level where at every moment multiple
influences tend to randomize the sequence of events.

From a genetic standpoint, it is fully accepted that "two individuals
of the same species develop according to a common design only insofar
as gene-controlled mechanisms reproduce the same conditions within
and around the embryo" [3] ; and no geneticist believes, with respect to
traits observed in one-egg twins, that symmetry and concordance are en-

Psychogenetic Studies of Twins 335

tirely owing to genotypic similarity or to the effect of single genes. The
significant influence of the embryonic environment on traits which are
not completely controlled by heredity has been demonstrated by Wright's
study of polydactyly [102], and the important effect of various combina-
tions of modifying genes on the expression of traits produced by a single
mutant gene is equally well established.

In other words, individual differences between one-egg twin partners
are no precise measure of environmentally produced variations, nor is it
likely that exact quantitative values will ever be assignable to the relative
contributions of genetic and nongenetic factors in the production of
these differences. It has even been suggested by Allen [3] that any esti-
mate of penetrance based on observations in one-egg twins should be
regarded "as a probable overestimate, or as the upper limit of the range
in which the true penetrance may lie."

Regardless of the fact that similar reservations apply to estimated
expectancy rates for two-egg twins (concordance as to a given type of
morbidity), it is apparent that an unrepresentative increase over the
average difference between dizygotic twin partners is no indication of the
exact contribution of genetic influences even in relatively comparable
environments. Two-egg twins are also closely related, so that it should
not be very difficult to find a few pairs displaying virtually the same
degree of concordance as one-egg twins. In fact, if in a study of in-
tellectual or motivational similarities the search for twins of either type
is restricted to pairs attending the same class in a certain college, or if
a comparison of biological health and survival values is made only in
complete same-sex pairs who have survived to the age of ninety-five years,
one should not be too surprised to obtain smaller intrapair differences in
two-egg rather than one-egg twins, at least once in a while. On the
whole, it is to be expected that observable differences between two-egg
twins of the same sex will somehow depend on the extent of dissimilarities
between their parents. In a random-mating human population, the given
intrapair differences will be more highly correlated with parental than
with grandparental differences [5].

With an understanding of the fact that generalized conclusions can-
not be drawn from observations made in single pairs or in an unrepre-
sentative series of pairs, the importance of adequate sampling procedures
with complete ascertainment of twin index cases (rather than pairs) in a
certain district or group of institutions becomes axiomatic. The sampling
methods to be used are essentially the same for twins and nontwins, but
the establishment of the twinning attribute in a given part of a popula-
tion requires a systematic screening procedure, including careful ex-
amination of official birth records wherever possible.

According to Allen [3], the most useful evidence of unbiased sampling

336 FRAXZ J. KALLMANX

in a twin study Is provided by an approximate agreement between a twin
sample and the parent population, either with respect to the proportion
of opposite-sex pairs or, when known, of one-egg pairs. Roughly speak-
ing, the population twin rate is 2 per cent, and the proportion of twins
from opposite-sex pairs is about one-third of all twins. Any series of
twins differing significantly from the parent population in either of
these two characteristics can be safely regarded as selected and un-
representative. It may be reemphasized, however, that the given statistics
are to be based on twin individuals rather than pairs, since even random
and representative samples may deviate considerably from the parent
population in statistical estimates based on pairs.

In the United States, the precise twin rate is 2.19 per cent of all
babies born since 1928. This rate is reduced to about 1.9 per cent by
excess twin mortality within the first year of life, whereas the 2 : 1 ratio
of same-sex to opposite-sex pairs observed at birth remains virtually un-
changed in all age groups. General mortality is higher in males than in
females, an increase that is apt to have some effect on the proportion of
male twins in an adult sample. However, the reduction of male twins is
not sufficiently pronounced to necessitate modification of the 1 : 1 sex
ratio in estimating the proportion of one-egg pairs by means of Wein-
berg's differential method [3].

After the first year of life, there is no significant difference between
the mortality rates of twins and nontwins, so that the proportion of twin
individuals in the population is assumed to remain nearly the same at all
ages. However, Allen has called attention to the fact that as pairs are
broken by mortality and migration, the number of intact pairs is reduced
at successive ages, thus making it necessary to analyze twin data in terms
of individuals rather than of twin pairs. Among people who survive to
an advanced age, the relative frequency of pairs represented by at least
one twin may be nearly twice as great as at birth.

Another point to be considered in the analysis of twin family samples,
in relation to specific pathological traits and their variations in different
periods of time or life as well as in different ethnic or socioeconomic
settings, is the need for consistently corrected morbidity risk figures (ex-
pectancy rates as obtained by the Weinberg method). In clinical in-
vestigations, expectancy rates are more valuable than the usual prev-
alence statistics favored by public health authorities. According to
Stromgren [84], the disease expectancy is "the risk of becoming ill during
one's lifetime, if one lives long enough to pass the period of risk55 (the
time during which the disease may develop ) .

Once again, the statistics describing such a sample are to be com-
puted from twin index cases (probands) rather than from twin pairs, if
some or many pairs are represented by a single index case, and if

Psycho genetic Studies of Twins 337

morbidity expectancy rates for various groups of siblings and co-twins
are to be compared within the sample (twin-family method). The same
method of computation is required for estimating penetrance and con-
cordance rates. In the latter case, however, it should not be overlooked
that the number of index cases from concordant pairs is to be halved
in order to correct for the twofold representation of concordant pairs in
the sample [3]. Since the concordance rate is usually understood to pro-
vide the directly established proportion of pairs with two affected
partners, if ascertainment of affected twins is complete for the popula-
tion, it is evident that the number of concordant pairs is one-half the
number of cases observed in these pairs.

In most instances, of course, differences between one-egg and two-
egg groups of twins will have the same statistical significance whether
evaluated in terms of concordance or morbidity expectancy. For the use
of concordance rates obtained under conditions of incomplete ascertain-
ment of twins affected by a pathological trait, various corrective formulas
have been devised by Allen [3].

In the analysis of normal personality variations in twin samples, the
best-known statistical technique employed in estimates of genetic com-
ponents is Holzinger's h2: the variance of the dizygotic twins minus
the variance of the monozygotic twins, divided by the variance of the
dizygotic twins [67]. In order to establish the significance of the h2
values, an F test may be used for the ratio of the dizygotic over the
monozygotic variance.

More recently, Gattell and associates [16] introduced a multiple
variance analysis design as a refined method for analyzing "dimensions
of personality which have been established by factor analytic investiga-
tions upon personality responses in rating data, questionnaire data, and
objective tests." The twelve primary personality factors measured were
obtained on the Junior Personality Questionnaire Test and included
three factors which provided evidence for predominantly genetic deter-
mination: general intelligence, cyclothymia vs. schizothymia, and ad-
venturous cyclothymia vs. submissiveness. Four factors assigned equal
roles to heredity and environment, although "heredity predominated be-
tween families33 (energetic conformity, dominance, socialized morale, and
impatient dominance). The predominantly environmentally determined
personality factors consisted of tender-mindedness, general neuroticism,
surgency-desurgency, will control, and somatic anxiety. A final report
on the results of this promising study has not yet been published.

As a general principle for the analysis of normal personality traits,
it may be worth mentioning that the given twin data should be expressed
in terms of varying degrees of intrapair similarity or dissimilarity, rather
than in terms of concordance or discordance. Twins may be concordant

338 FRANZ J. KALLMAXX

or discordant as to rheumatic heart disease, but not as to the normal
shape or color of their hearts.

PROCEDURAL LIMITATIONS AND ADVANTAGES

Regarding the procedural potentialities of twin studies, research
workers in psychological genetics, not to mention their critics5 would
do well to remember that, like any other scientific procedure, the twin-
study method has its limitations and its very specific advantages. It is a
mistake to overstrain the merits of the method by letting it bear the
burden of proof in extravagant attempts to measure multitudinous
variables in "the dilemma of mind-body dualism" [65],

It is equally inappropriate to belittle the value of twin studies either
because of some inherent imperfections in their applicability, or because
of an unexpected lack of success in overexpanded investigations con-
cerned with "concepts of absolute or ultimate causes'3 [65]. In the
humble words of Carl Lotus Becker, "the significance of man is that he is
insignificant and is aware of it."

In general terms, the limitations of the twin-study method can be
placed in three categories:

1 . Imperfections of the research species.

2. Imperfections of pluridisciplinary research workers, research
methods, and research teams.

3. Imperfections of quantifying methods for measuring meaningful
personality differences in genetically similar or dissimilar phenotypes.

As to the first group of limitations, it cannot be helped that, like
every other human research subject, human twins enjoy more sacred
rights, a longer Me span, and more intricate systems of organization and
regulation than any species of laboratory animal. Of course, they can-
not be kept in cages, nor can they be separated before they are born.
Forced to exist in crowded quarters during important stages of embryonic
development, they may carry some genes which are sensitive to asym-
metrical cytoplasmic influences in this prenatal period [22]. Also, they
are prone to prematurity [51] and birth trauma [4] and seem to have a
preference for "non-white55 mothers, who do not belong to "the lower
socio-economic segments of the population" [61].

Following the ordeal of being born, twins enter a world in which
parents rear their own children, thereby depriving similar as well as dis-
similar twin partners of the chance of benefiting from, or being observed
in, entirely different cultures. Instead, some twins will be subject to
superstitious beliefs, fierce parental pride or bewilderment, the hazards of
educational laxity, easily mistaken identity, or being thought of in terms

Psychogenetic Studies of Twins 339

of behavioral opposites, and the potential effects of a modified form of
sibling rivalry.

For instance, the marked similarity of school performance ratings
in one-egg twins has been explained by Husen [41] on the basis of such
incidental factors as going to and from school together, or being mis-
taken for each other by teachers. Zazzo [103] ascribed a considerable
part of an observed IQ deficit (9 IQ units) in a European twin popula-
tion to a language retardation resulting from the twins' preference for
using a "secret language" with concomitant social isolation.

In fact, Bauer [11] believes that one-egg twins can be regarded as
psychologically incomplete individuals "sharing an ego53 and projecting
each other's inner life in their manifest behavior. On the other hand,
Burlingham [14] has expressed the opinion that twins have "a more
acute rivalry to cope with than is the case with ordinary siblings.33 Com-
pared with the rivalry among siblings, that between twins was found by
this investigator to start at an earlier age, to be more pronounced "be-
cause33 of the necessity of competing on an equal footing, and to cul-
minate more frequently in early mutual death wishes.

Other potentially disadvantageous aspects of twin development have
been seen in a weakened relationship with the parents, as a corollary of
an intensified identification process between the twins, in a bewildered
parental attitude toward two children who are so alike, and by erroneous
analogy with freemartins in cattle, in the purported sterility of one of
identical twin brothers [50]. According to Burlingham's theory the
identification mechanism tends to preserve a marked degree of similarity
between twin partners that would otherwise gradually yield to significant
differences in behavior.

Fortunately, most of these interpretive inferences regarding the pre-
cariousness of a twin's childhood, adjustment, or reproductive capacity
are far too gloomy. Once a twin has survived his first year of life without
evidence of organic damage, he is virtually certain to be undistinguished
from single-born individuals, even to the extent of having a complete
ego of his own. There is no evidence of premature babies being more
likely than full-term infants to develop a psychosis, nor are there any
statistical indications that infections, emotional disturbances, or other
tangible disabilities are more prevalent in twins than in the general
population.

On the contrary, twins are known to vary as much in their per-
sonalities, intellectual abilities, and stress symptom thresholds as do single-
born people, and there is no reason to believe that they are less healthy,
less longevous, or less selective in regard to their own potential formula
of adjustment. In childhood they seem to be as capable as many other
children of working out fairly adequate ways of dealing with difficulties

340 FRANZ J. KALLMANN

in their home milieu and interpersonal relationships [45]. As adults they
show the expected variations in marital and reproductive histories, and
when tested psychometrically in far-advanced age, they produce mean
scores which are entirely "within the limits of normal expectation" [50],

Since even Burlingham reported that the early differentiation of
roles, said to divide one-egg twins into one active and one passive partner,
was found to be "determined by the bodily strength of the children and
to change according to changes in their relative health and development"
[14], it cannot be assumed "that with regard to intelligence or general
vitality or any other aspect of biological development, a twin derived
from only one-half of a fertilized ovum might tend to be inferior either
to a two-egg twin or to the average single-born person" [50].

It is reasonable to say, therefore, that the main disadvantage of twins
as research specimens lies in certain imperfections which characterize
human beings, human societies, and human vicissitudes in general. By
the same token, the willingness of twins to serve as research subjects in a
cooperative spirit somehow depends on establishing and maintaining
personalized relations with them. This requirement calls for a substantial
degree of empathy, humbleness, and sincerity on the part of the in-
vestigator.

The common denominator for the second set of twin-study limitations
— imperfections consistent with pluridisciplinary research workers, re-
search methods, and research teams — is the fact that twin researchers are
human, too. They cannot expect to be either more longevous or much
more versatile than their research subjects, and they always risk frustra-
tion in the conduct of studies which cut across the customary borders of
individual disciplines.

Since they cannot hope to qualify as experts in every discipline deal-
ing with the structural, physiological, or psychological aspects of per-
sonality development in normal and pathological constellations, gemel-
lologists must learn to be satisfied with fractional answers to pluri-
dimensional problems. On this level, however, they are easily misled into
describing observed temporary and perhaps reversible dissimilarities be-
tween twin partners in the antithetical setting of absolute dichotomies,
or in mystifying terms borrowed from transcendental schools of thought.

Ideally speaking, individual twin research workers, depending on
their professional qualifications, should have little trouble delineating
their tasks according to whether the variations studied fall into the
normal or pathological ranges of variability. Unfortunately, the dividing
lines are seldom clearly drawn, and there are not too many research
workers who are willing to anticipate limitations in their investigative
capacities.

The alternative is the formation of interdisciplinary research teams

Psycho genetic Studies of Twins 341

for conducting twin-study projects which are broad as well as continuous
in a longitudinal scheme. However, apart from being expensive and dif-
ficult to maintain over a prolonged period of time, such an organization
may prove susceptible to professional rivalries and the introjection of
incompatible biases and predicative jargons. So long as there is still much
intradisciplinary disunity in most of the behavioral sciences, it would
seem advisable to bear in mind that there is a point of diminishing re-
turns in the prospects of pluridisciplinary projects.

On the genetic side, two divergent philosophies have emerged — the
statistical (static) and the physiological (dynamic) points of view. The
statistical philosophy has been described as hyperatomism and hyper-
selectionism [37] and is suspected of interpreting every generalized set of
facts by the introduction of more and more units for statistical treatment.
This pattern may lead to serious consequences by requiring "astronomical
numbers of modifiers and a similar number of tiny but specific adapta-
tions." In order to explain all phenomena which appear to be gene-
controlled, more and more genes are introduced in the form of modifier
systems built up by selection.

The dynamic approach is preferred by Goldschmidt's school, as well
as the writer. Although it accepts the basically statistical tenets of
genetics, the main objective is seen in an understanding of behavioral
phenomena in terms of gene-specific molecular processes and develop-
mental systems, with all their interaction, embryonic regulation, and inte-
gration. In this frame of reference, the concept of Mendelian heredity
(with or without simple segregation ratios) becomes more or less
synonymous with "chromosomal heredity." Pertinent environmental
factors which mold, and the formative elements which secure behavioral
malleability on the human level are viewed as "end-products of the same
evolutionary process," and are likened to "the two sides of a coin,
defying analysis as independent variables" [45].

On the psychological side, there is an even more perplexing division
into schools, each of which rejects some of the fundamental standards of
classification accepted by the others [82]. The emotional tone of this dis-
pute has been compared by Slater to that which was rampant in "the
days of debate between allopaths and homoeopaths over a century ago,"
This country has also had its share of the spurious nature-nurture battle
for supremacy, the unfortunate effects of which have been commented
upon by many writers.

Especially relevant here are Skinner's reflections on science as "a
continuous and often a disorderly and accidental process" and on the
scientist as "the product of a unique history" who may be "more con-
cerned with his success as a scientist than with his subject matter" and
may therefore aspire to assume "the role of a roving ambassador" [78].

342 FRANZ J. KALLMANN

In this context, psychiatry is presented as "a field in which behavior is
customarily described . . . indirectly.35 Hence, psychologists are cau-
tioned to put "the older statistical and theoretical techniques in their
proper perspective55 and awaken to the possibility that there may be
direct observation of behavioral processes.

Agreeing with Rogers's notion that "science is not an impersonal
something, but simply a person living subjectively another phase of
himself . . . and that the knowledge gained through scientific method
... is a matter of subjective choice dependent upon the values which
have personal meaning for me" [71], Skinner is opposed to any attempt
"to fit all scientists into a single mold." From a gemellological stand-
point [45] 5 it is also imperative for workers in the behavioral sciences to
discontinue the practice of using a two-valued system of conceptualiza-
tion in dealing with problems of personality development.

So long as belief in heredity as an essential determinant of variable
behavior patterns is equated by some workers with a fatalistic distrust of
man's perfectibility, it will be difficult to form pluridisciplinary research
teams unhampered by rigidly codified schemes of dichotomous absolutes.
Twin studies foster the hope that environmental variables are not the
only ones which can be controlled by man. However, research workers
are required who are not afraid of delving into the dimly lit strata of
man's bipolar existence. Only as a team learning to avail itself of the
opportunity to break new ground with the aid of new research methods,
can these wrorkers come to understand basic behavior patterns in terms of
physiochemical or molecular processes powered by genie elements.

Of course, the research tools employed in the conduct of comparative
twin studies cannot possibly be less imperfect than the research workers
using them. Since these studies are largely concerned with problems at
the very beginning of an intricate chain of cause and effect [82], their
usefulness depends on an unbiased collection of clinical, demographic,
and psychometric data. Therefore, it follows that most of the im-
perfections of pluridisciplinary twin research workers are also inherent in
their research methods. Relatively few workers in the behavioral sciences
are able to accept the virtual inseparability of genetic and nongenetic
components of personality and the fact that most test devices for measur-
ing personality differences have proved refractory to standardization
[9,18,47].

Thus, for the sake of respectability a gene-specific effect on human
behavior is acknowledged by some euphemizing analogy, or by keeping
it in what may be called the "etcetera" category of contemporary psy-
chology and psychiatry. The tendency to regard accounts of gene-con-
trolled phenomena as the work of a devil's advocate persists [45]. The
temptation to avoid labeling genetic factors as such is most likely to
arise when psychometric attempts are made to assign quantitative values

Psycho genetic Studies of Twins 343

to the relative contributions of genetic and environmental factors in pro-
ducing individual differences, or whenever it becomes necessary to dis-
tinguish between cause and motivation in human behavior.

It is ironic that the same factors producing imperfections in the twin-
study method should also be listed among the principles specified as
procedural advantages. Nevertheless, both the humanness of twins as
research specimens, and the need for employing interdisciplinary re-
search teams are assets as well as disadvantages of the method. Rabbits
are out of place when it comes to differentiating between the symptoma-
tologies of schizophrenia and manic-depressive psychosis through the be-
havior of monozygotic individuals; and the best way for members of
different disciplines to learn to respect each other's work is to give them
a chance to work together.

Compared with other techniques for studying normal or abnormal
variations in human subjects, families, and populations, the twin-study
method has the following specific advantages:

1. Except for traits which are peculiar to twins or significantly
altered in twins, the method constitutes an excellent sampling procedure
for the investigation of variations displayed by different genotypes in
a controlled environment, or by a constant genotype under the influence
of different environmental conditions.

2. In the study of traits which require close personal contact with
the research subject and access to all strata of a given population, the
method provides an inconspicuous approach to families whose private
affairs might not otherwise be open to study.

3. Likewise, the method facilitates the conduct of combined cross-
sectional and longitudinal investigations. Such investigations are in-
dispensable in pathological conditions where information is needed not
only as to the selection pressures bearing upon affected persons, but also
as to variations related to age of onset, duration, or severity of clinical
symptoms, expected distribution between the sexes, or the differential
aspects of reproductivity, responsiveness to treatment, and other survival
values [43].

4. In conjunction with the statistical principles of the proband and
sibling methods (Weinberg), a twin study covering an entire district
or state represents the most economical substitute for a total population
survey requiring personal interviews and the application of controlled
test procedures.

SERIAL TWIN DATA ON INTELLECTUAL AND PERSONALITY
VARIATIONS

In reviewing the part twin studies have played in analyzing variations
in intellectual abilities and personality potentials, no attempt will be

344 FRANZ J. KALLMANN

made to achieve bibliographical completeness. The report is focused
on the results of fairly recent investigations in which a sizable series of
twins was used. Pertinent data have been considered primarily from a
genetic viewpoint on the clinical side, and from the standpoint of their
potential psychological significance on the genetic side. The role of non-
genetic influences on the psychometric, psychodynamic, and pathoplastic
aspects of personality differentiation is dealt with elsewhere in this
volume. Needless to say, full appreciation of these influences Is not in-
dicative of loyalty to any one school of training.

General understanding of the interaction of heredity and environ-
ment is greatly enhanced by a dynamic concept of constitutional, in-
tellectual, and characterological phenomena based on the solid founda-
tions of genetic principles [45]. Since the organism is both active and re-
active, the importance of genie elements in the organization of behavior
patterns rests on the interdependence of organic structure and psy-
chological function throughout the life of the individual. There is no
behavior without an organism, no organism without a genotype, and no
physiological adaptedness without continuous and fully integrated gene
activity.

Of course, in order to maintain his present evolutionary level, man
must be both conditionable by culture and impressible by education. The
ability to learn from others and to profit from experience is determined
by the genotype; cultural values and opportunities have to be acquired
by each individual through communication with his group. Broadly
formulated, then, a person's phenotype may be defined as the visible
expression of his malleability by environmental influences, and his geno-
type as determining his norm of reaction to the total range of possible
environments during his lifetime. The implication here is that every
gene-controlled mode of activity requires an operational area in which
to unfold [9, 23,47, 65].

In the area of normal personality variations, the earliest twin studies
were limited to a descriptive account of the histories of interesting twin
pairs, without looking at the backdrop of the total genetic and en-
vironmental variation observed in a population. More recent data have
been based on serial studies in which the development and performance
of each twin were compared with those of his partner. Early pilot studies
were those by Galton [32] in 1883 and Thorndike [89] in 1905. They
were followed by the work of Von Verschuer [92], Lange [56], Wing-
field [101], Herrman and Hogben [40], Graewe [38], and Gedda [33]
in Europe, and by that of Rosanoff and Orr [72], Gesell [35], Merriman
[64], Lauterbach [57], Newman and associates [67], and Burks [13] in
the United States.

On the heels of progress made in the procedures of clinical classifica-

Psychogenetic Studies of Twins 345

tion and zygosity determination, psychiatric twin data of more than
historical interest were presented by Luxenburger [62] in 1930, and by
Rosanoff et al. [74] in 1934. Our New York State studies were organized
in 1936 [43, 8]. They were matched by Essen-Moller [26] in Sweden
(1941) and by Slater [80] in England (1951) with series of fairly com-
parable size.

In Galton's study [32], subjective estimates of intelligence and
zygosity were used. Thus he observed that ordinary environmental dif-
ferences w^ere not sufficient to make "similar" twins unlike, w^hereas
"dissimilar55 twins were not found to become more alike under the in-
fluence of similar surroundings.

Thorndike's series [89] consisted of 50 pairs unclassified according
to zygosity. His working hypothesis was that higher intrapair correlations
would be obtained in "trained53 rather than in "untrained" functions,
if the training itself were responsible for similarities in scholastic achieve-
ments and comparable intellectual functions. Evidence for a primarily
genetic determination of variations in mental abilities was seen in the
finding that resemblances between twins changed neither with age nor
with training.

In the studies of Merriman [64] and Lauterbach [57], data obtained
by the Stanford-Binet, Army Beta, and National Intelligence Test, as
wrell as intelligence estimates by teachers, were evaluated in 100 and 200
pairs. The observed resemblance in IQ was of the same order of magni-
tude for male pairs (0.877 ± 0.30) and female pairs (0.857 ± 0.029),
and older twins were not found to be more alike than younger ones.
Wingfield [101] introduced various procedural refinements (using 102
pairs in the age group seven to fifteen) , but obtained essentially the same
results as the earlier investigators.

Herrman and Hogben [40] compared only very similar one-egg twins
(65 pairs) with very dissimilar two-egg twins of the same sex (96 pairs),
employing the Otis Advanced (Form A). Apparently, the observed one-
egg correlation (0.86 ± 0.04) was somewhat too high, and that for the
same-sexed two-egg series too low.

Like Newman, Freeman, and Holzinger [67], Burks [13] was
especially interested in the development of one-egg twins separated in
early childhood. Her observations on four pairs of this type were in-
terpreted as indicating the significance of both genetic and nongenetic
factors in shaping the life histories of genetically alike partners.

In the Chicago study, 100 nonseparated, same-sexed pairs (50 one-
egg, 50 two-egg) of school age (eight to eighteen years) were compared
with 19 one-egg sets (aged eleven to fifty-nine) who had been separated
early in life. The Stanford-Binet correlation for one-egg twins reared
apart was 0.77, about midway between those for two-egg twins reared

346 FRANZ J. KALLMANN

together (0.63) and one-egg twins reared together (0.88). The In-
vestigators concluded that only extreme environmental differences tend
to have an appreciable effect upon intelligence.

Generally corroborant data were reported by Thurstone et ai [90]
for a sample of adolescent pairs (48 one-egg, 55 two-egg), by Baroff
[10] for a series of 40 one-egg and two-egg pairs with high-grade mental
retardation, and by Feingold [28] and the writer [46] for a series of 127
same-sexed senescent pairs (mean age sixty-nine and seven-tenths years)
studied longitudinally over a period of eight years.

The extensive batter}' of Thurstone and associates consisted of tests
chiefly measuring primary mental abilities, personality, and psychomotor
function. The analysis raised almost as many questions as it answered,
because of inconsistencies in motor function scores for the two hands
and the failure of reasoning and mathematical faculties to distinguish
the two zygosity groups. Nevertheless, the study substantiated the as-
sumption of an important genetic component in those abilities dif-
ferentiating one-egg and two-egg pairs. Especially on some of the visual,
verbal, and motor tests, two-egg twins displayed marked intrapair dif-
ferences with significantly increased frequency. The decisiveness of this
finding was confirmed by the preliminary data of Vandenberg [91] re-
ported in 1955.

In Baroff's investigation of intelligence as measured by mental age,
one-egg twins proved to be significantly more similar than two-egg twins,
despite the fact that this institutional series included only pairs con-
cordant as to mental retardation. In the one-egg group, the degree
of similarity in mental age remained unaltered by the duration of
kistitutionalization (relatively constant environment), whereas two-
egg twins showed increasing disparities. In the author's opinion, genet-
ically unlike persons in a similar environment are likely to become in-
creasingly dissimilar in the symptomatology of an inherited type of
mental defect.

Feingold's impressive data were collected in conjunction with the
senescent twin population study organized by this writer and his as-
sociates in 1945 [28, 46, 50]. The main purpose of the project was to
investigate intrafamily variations in aging patterns in both cross-sectional
and longitudinal directions. The total sample consisted of 2,536 senescent
twin index cases in New York State (sixty years of age and over), in-
cluding a series of 1,557 index pairs whose zygosity was sufficiently
established to be useful for comparative longevity analysis. The total
number of one-egg and two-egg pairs, observed with respect to health
status, intellectual performance, and length of life, approximated a
1:2 ratio (518:1039) and was in accordance with statistical expecta-
tion. At the beginning of 1956, after eleven years of observation, 516

Psychogenetic Studies of Twins 347

twin subjects were still alive, including 179 pairs where both members
survived.

The psychometric study was planned in such a way as to provide
comparable test scores of same-sexed twins on a longitudinal basis. The
240 test cases chosen for this purpose in 1947 had to meet the require-
ments of being white, literate, native-born, noninstitutionalized, and
apparently free of mental and physical illness. Of this sample, 36 com-
plete pairs and 7 single survivors were retested with the same battery

60r~

E30
§20

O
CL

10

13 Monozygofic
D Dizygof/c

1947

60 h

£ 30

a)

t? 20

c£
10

1955

TEST Vocabulary Digit Block Similarities Tapping Digit

symbol design span

FIG. 1. Comparative mean intrapair differences in test scores (1947 and 1955).

in 1955, after a mean interval of 7.8 years. The battery consisted of four
subtests taken from the Wechsler-Bellevue Scale I (digit span, similarities,
block design, digit symbol), the vocabulary list of the Stanford-Binet
1916, and a paper-and-pencil tapping test. At the time of the retest in
1955, the age of the survivors ranged from sixty-eight to eighty-seven
years, with a mean age of seventy-four and one-half years.

The results of the first test round showed that the mean intrapair
differences in test scores measuring various intellectual abilities were con-
sistently smaller in one-egg than in two-egg pairs (Fig. 1). The dif-
ference between the two zygosity groups was significant at the .01 level
of confidence for the vocabulary, digit symbol, and tapping tests. In

348

FRANZ J. KALLMANN

accordance with Zazzo's finding [103] that the difference between the in-
telligence quotients of twins and nontwins tends to decrease with in-
creasing chronological age, no evidence was obtained in support of a dif-
ference between the test performances of aging twins (one-egg or two-
egg) and comparable single-born persons. There was a difference be-
tween male and female test scores, indicating that certain intellectual
changes in the period of senescence are observable in males at an earlier
age than in females. On the whole, the test data clearly revealed that
gene-specific intellectual differences persist into a well-advanced age.

Similarities

Longitudinal
Cross -sectional

Block design

Digit symbol

60 65 70 75

Aqe in years

FIG. 2. Trends of intellectual decline in senescence (longitudinal and cross-sectional
test data).

Although the retest series [46] was numerically too small to show
statistically significant differences between zygosity groups, it was still
apparent in five out of six tests that the mean intrapair differences tend
to be greater in two-egg than in one-egg pairs. The digit span test data
were the exception.

As to the longitudinal trends revealed by the testable survivors, the
test results showed a consistent although slight decrement in intellectual
abilities during senescence. By and large, this finding was in agreement
with the trend observed in cross-sectional investigations, but the slope of
the decline in the longitudinal study (Fig. 2) was smaller than that ex-
pected on the basis of survey data.

Psychogenetic Studies of Twins 349

Another interesting observation was that the retested twin pairs
scored higher on the original tests than did the total sample tested
previously. This disparity seemed to indicate a relationship between test
score level and survival potential. However, without corroboration by
data from larger samples, the implications of this hypothesis could not be
regarded as conclusive. The same reservation had to be made for the
finding that two-egg twin partners who are most similar in test scores
in the senescent period may have the best chances of surviving together.

TABLE 1. BIENNIAL MEAN INTRAPAIR LIFE SPAN DIFFERENCES IN SAME-SEX TWIN
PAIRS OVER AGE 60 (BOTH DECEASED)

Intrapair life span differences

Year of

Number of

expressed in months

analysis

index pairs

Male

Female

Total

1948

32

47.6

29.4

36.9

1950

68

42.9

31.2

36.7

One-egg pairs

1952

76

40.7*

30.7*

35.7*

1954

78

40.7*

31.6*

36.0*

1956f

104

49.9

33.9*

41.9*

1948

36

89.1

61.3

78.3

1950

70

79.1

63.2

71.8

Two-egg pairs }

1952

86

79.1*

69.5*

73.7*

1954

102

69.5*

79.1*

74.6*

1956f

110

69.2

75.3*

72.5*

* Significant at 1 per cent level.

f Preliminary data.

± All opposite-sex pairs over age 60:106.0 months.

How genetic factors determine the ordinary length of life and other
general health and survival values has been shown by comparing the
life spans of those twin pairs where both partners died of verified natural
causes after the age of sixty years (Table 1). In all the biennial
estimates made since the beginning of the study, the mean intrapair life
span difference has been smaller in one-egg than in two-egg pairs.
The present total mean difference (1956) varies from 41.9 months in
the one-egg group to 72.5 months in the two-egg group of the same sex.
The differences between the two zygosity groups have been statistically
significant (p — .01) in the last three analyses, although there is an ex-
pected disparity between the contributions made by the two sexes to the
total difference., probably owing to the shorter life span of the male.
The present difference between the mean intrapair life spans of the two

350 FRANZ J. KALLMANN

male groups is close to 20 months, but It does not reach the level of
statistical significance if only males are considered.

In general, as to normal personality variation it can be said that
gene-specific derivations range from physical, coordinative, physiognomic,
and temperamental characteristics to intellectual abilities, affective
regulations, and special talents [39, 43, 90, 91]. In between are sex
maturation patterns, variations in antibody production, the capacity for
longevity, and the ingredients for sustained tolerance of physiological or
psychological stress, a highly essential prerequisite for a well-balanced
personality [45, 86]. Except for one-egg twins, it is apparent that each
individual has his own threshold of adaptability to different types of
stress, and his own pattern of stress symptom formation.

Consistent similarity in the composition of these personality com-
ponents is not observed in the absence of genotypic identicalness. Two-
egg twins of the same sex tend to differ as much in their personalities as
any siblings reared together or apart. Only one-egg twins retain basic
similarities in appearance and general personality traits despite pro-
nounced differences in life experience.

This principle is not refuted by the fact that a spiral-like develop-
ment toward marked behavioral dissimilarity (chronic alcoholism, de-
linquency, suicide) may sometimes result from a seemingly insignificant
difference in the original adjustive patterns of one-egg twins [43, 45, 81].
Extreme disparities of this kind are the exception rather than the rule.

SERIAL TWIN DATA ON PSYCHOPATHOLOGICAL VARIATIONS

In considering the contributions made in the area of psychopatho-
logical variations by means of serial twin data, it may be helpful to bear
the following points in mind : from a genetic viewpoint, the dividing line
between a normal state of adjustment and those minor forms of ill health
commonly referred to as psychoneurotic is not regarded as static, nor as
less vaguely defined than that between normal and subnormal intelligence.
A deviant behavior pattern is not presumed to be the result of a simple
genotype-phenotype interplay, reducible to an aggregate of well-de-
lineated causes and effects, nor is it merely thought of as the concomitant
of a fixed congenital aberration, or as a self -limiting error in homeostasis,
or as just an unfortunate episode in adjustment [68, 96].

Genetically, human behavior of any variety is viewed as an extremely
complex and continuous chain of events in the individual's adaptive
history. It is axiomatic, of course, that even the finest genetic endow-
ment can go astray, either because of an unusual combination of ad-
verse circumstances (intrinsic or extrinsic) or because of prolonged
abuse.

Psycho genetic Studies of Twins 351

An example is the tendency to suicidal acts, which was long thought
to be due either to an inherited degree of unfitness in general personality
structure or to a special reaction type distinguished by introjective
aggressiveness. Neither theory, however, has been supported by our data
on a series of 18 one-egg and 21 two-egg pairs of twins, one of whom
had committed suicide [43]. With but one exception, all have remained
discordant. In short, suicide is one of the few phenomena unlikely to
occur in both twins even under similar conditions of maladjustment and
privation.

The high-tension state released by a suicidal mechanism, in the form
of a self-destructive trigger reaction to adverse life conditions (com-
pulsive or twilight-state type of short-circuit reaction under stress),
apparently depends on unusual and not easily duplicated constellations of
motivational factors. Although two twin partners may both commit
suicide, it will only be by chance and without direct relation to each
other. Thus, concordance will be extremely rare even in one-egg twins.

As to psychoneurotic reaction potentials (outside the field of crim-
inality), Eysenck and Prell [27] had the courage to join the small group
of investigators who availed themselves of the opportunities afforded by
the twin-study method. In Hne with their findings in a series of 25 one-
egg and 25 two-egg pairs, they classified "the neurotic personality factor"
as a biological and largely gene-specific entity, estimating the genetic
contribution to this "neurotic unit predisposition55 as 80 per cent.

Not quite so specific is Slater's [80, 82] interpretation of the neurotic
symptoms, observed in a series of 9 one-egg and 43 two-egg pairs, as
exaggerations of polygenically determined personality variants, less closely
related to a given type of stress than to the basic personality. Despite
"almost identical personality,55 seven of the nine one-egg pairs failed
to present concordant psychoneurotic histories, as against 15 concordant
pairs in the two-egg group. Therefore, critical deviations in a person's
career were assumed to be due to relatively chance occurrences, such
as the personality of the chosen marital partner. "One twin might suffer
a mischance which would lead to a vicious circle of ill-health, social
failure, hardship, discouragement and increased ill-health, while the other
totally escaped.55 According to this theory, there are graded constitutional
vulnerabilities in more than one dimension, so that "the man who breaks
down with a neurotic illness is likely to be handicapped not by one con-
stitutional weakness of severe degree, but with a number of minor
weaknesses."

However, Shields's study [76] of 62 same-sexed pairs between the
ages of twelve and fifteen years (36 one-egg, 26 two-egg) provided
evidence for one-egg twins (69 per cent) being twice as likely as two-egg
twins (31 per cent) to have the same degree of adjustive difficulty. With

352 FRANZ J. KALLMANN

each child rated "on a four-point scale of psychiatric maladjustment/5
twins of either zygosity had no higher incidence of neurotic adjustment
problems than single-born controls, but male twins and nontwins far ex-
ceeded their female counterparts in presenting some difficulty in adjust-
ment. Little more than one-half of the total group of English school
children investigated (four South London areas) were classified as non-
neurotic. Nongenetic explanations for the observed differences between
one-egg and two-egg twins were rejected, perhaps somewhat summarily,
especially in regard to the frequent similarities in type and severity of
neurotic behavior patterns.

One of the highest one-egg concordance rates reported has been that
for homosexual behavior in the adult male [43], although all concordant
twin partners in this series denied any mutuality in overt sex relations.
Nevertheless, 44 one-egg pairs yielded a nearly perfect concordance rate,
with the index cases standing at least midway on the homosexuality
scale applied, and with pronounced similarity in the role taken by twin
partners in their individual sex activities. In the two-egg group (51
pairs), nearly 60 per cent of the co-twins of predominantly or exclusively
homosexual index cases showed no evidence of overt homosexual behavior
at any age, and only 11.5 per cent were given homosexuality ratings of
five or six on Kinsey's scheme. The likeliest genetic explanation for these
findings would seem to be a gene-controlled disarrangement in the
balance between male and female maturation patterns, resulting in a
shift toward an alternative minus variant in the integrative process of
psychosexual maturation.

Another condition with a well-established one-egg concordance rate
of close to 100 per cent (6 pairs) is an entirely different defect of more
obvious organicity, namely, mongolism, the relationship of which to
maternal age is regarded as fully substantiated [7]. Since the correspond-
ing rate for two-egg co-twins (23 pairs) does not seem to exceed that
of their later-born siblings (approximately 4 per cent), the search for the
etiological factor in mongolism has been narrowed down by twin data
to a more or less permanent change in the mother's endocrine or repro-
ductive system. Apparently, the noxious influence during a mongoloid
pregnancy is not transient, but acts on a genetically predisposed embryo,
or upon the ovum, or upon the embryo before the earliest stage when
twinning occurs by division.

Serial twin studies have also aided in investigating the etiology of two
other organic syndromes, cerebral palsy [4, 52, 89] and convulsive
disease [17, 58, 59, 60]. As to the former condition, the data of Allen
(60 twin cases) and Thums (90 pairs) have indicated that twins are
rarely concordant. There is a high rate of stillbirth or neonatal death in
the co-twins of cerebral palsy cases, but no evidence for a specific genetic

Psychogenetic Studies of Twins 353

susceptibility to prenatal or natal injury. Apparently, many twins with
cerebral palsy are suvivors of adversities which proved fatal to their
twin partners. The circumstances most likely to affect both twins of a
pair include nonspecific maternal and genetic factors, as well as pre-
maturity per se, but probably not mechanical trauma during birth.

The most extensive analysis of epileptic twin pairs (30 one-egg, 130
two-egg) is that of Conrad [17], with concordance rates of 66.6 and
3.1 per cent, respectively. Unless plainly repudiated, the results of this
study represent strong evidence for the genetic origin of true convulsive
disease.

According to Lennox et al. [58, 59, 60], cerebral dysrhythmia is an
electroencephalographic expression of the epileptic genotype, assumed to
be the result of a dominant gene by the Boston group, and of polygenic
factors by Alstrom [43]. In epileptic twins, 25 per cent of two-egg and
100 per cent of one-egg pairs have been found to be equally dysrhythmic,
despite marked dissimilarities in clinical symptoms.

In the area of criminal behavior [53, 56, 74, 80, 85], there is still
an emphatic need for well-planned cross-sectional and longitudinal twin
data. Based on the findings of Kranz, Lange, and others [43], criminality
concordance rates vary only from 14 per cent in opposite-sexed pairs to
54 and 66 per cent in same-sexed two-egg and one-egg pairs, respec-
tively. This distribution indicates that both family milieu and basic per-
sonality traits play important parts in shaping the habitual criminal.
Therefore, the trend toward similar criminal behavior in two-egg pairs
may stem largely from the effect of unfavorable environmental influences.

Measured by the same yardstick, concordance in one-egg pairs can
often be expected to extend to specific personality features likely to lead
to a criminal career (brutality, ruthlessness, predatoriness, irresponsi-
bility), rather than to the kind of crime perpetrated. With criminality
itself determined in many cases by constellational circumstances, dis-
cordance may occur even in one-egg pairs, where one partner "manages
to keep within the bounds of law, while the other having once taken a
criminal step remains outside the law and does not find his way
back53 [80].

As to behavior disorders that are not sufficiently explained on a
situational or experiential basis, the list of conditions for which detailed
twin data are now available is headed by the schizophrenic and manic-
depressive types of psychosis [43, 45]. Since these two disorders do not
occur interchangeably in the same twin pairs, they are assumed to be
genotypically specific. The potentialities for a cyclic psychosis are prob-
ably associated with a subtle disturbance in a neurohormonal control
mechanism which ordinarily protects a person from having harmful
extremes of emotional responses. The concordance rates of two-egg and

354 FRANZ J. KALLMANN

one-egg twins (52 and 23 pairs, respectively) vary from 26.3 to 95.7
per cent.

Although the tendency to exceed the normal range of mood vacilla-
tion apparently requires the imitative effect of a single dominant gene
(with a tendency to incomplete penetrance), the metabolic deficiency in
a potentially schizophrenic person seems to be the result of a recessive
unit factor. Varying clinical expressions of the disordered behavior
pattern associated with the ensuing type of vulnerability to stressful ex-
periences are probably produced by a number of modifying genes. Gen-
erally speaking, these variations depend on the type and degree of con-
stitutional defense reactions that can be mobilized against the main
biochemical (enzymatic) dysfunction. The concordance rates for two-
egg and one-egg twins (based on a series of 953 pairs) are 14.5 and
86.2 per cent, respectively.

Involutions! melancholia and other nonperiodic forms of depressive
behavior in the involutional and senile periods have been shown by our
twin data (62 one-egg, 142 two-egg pairs) to be unrelated to the manic-
depressive group of disorders. There is an indirect link with the schizo-
phrenic genotype through certain forms of emotional instability char-
acteristic of schizoid personality traits. Other symptoms of maladjustment
in the senescent period may arise either from gene-specific metabolic
dysfunctions peculiar to the senium, or from graded differences in gen-
eral health and survival values.

Twin studies have helped materially to focus attention on numerous
obscurities in the etiology of all these disorders. Growing insight into the
cellular, structural, and metabolic aspects of personality organization will
gradually unfold a keener and more profound understanding of human
behavior.

PROSPECTS FOR FUTURE TWIN RESEARCH

Although the research data gathered by means of twin studies are
invaluable, a great deal of work has yet to be done, particularly in the
behavioral sciences. Admittedly, progress in psychogenetics has been
slow, and it may not be much accelerated in the near future.

Apart from a long delay in developing biologically oriented defini-
tions and classifications, only relatively few research organizations have
specialized in this area. Longitudinal twin family investigations are time-
consuming, expensive, and often destined to be narrow in scope. In many
instances, they may prove only that gene-specific determiners are essen-
tial in the etiology of a normal or deviant behavior pattern [6]. What
they may never be able to explain, however, without considerable help
from other disciplines, are the basic questions as to which genetic factors

Psychogenetic Studies of Twins 355

and how many are involved, what their biochemical actions are, and
how they interact with other genetic factors and with the environment.

Here a word of caution to workers in genetics seems indicated in
regard to the current tendency (Allen) to assume the operation of a
gene-controlled variation whenever some anatomical or chemical phe-
nomenon is found in a population giving evidence of a behavioral devia-
tion. With each new report of a possible organic correlate of a certain
type of mental disorder, it would be well to bear in mind that should
one of these findings stand up under scrutiny, it may turn out to be the
consequence of a patient's disturbed behavior, rather than its gene-
specific cause.

Of course, there is a tendency to oversimplify on the other side, too.
What is sometimes overlooked in the formulation of purely psycho-
dynamic theories is the fact that man is selective in the development of
his own formula of adjustment. For instance, there is no simple relation-
ship between a good home and normal behavior, any more than between
a poor home and mental disorder. In the absence of genotypic identical-
ness, even pronounced similarities in physical and cultural environments,
including having the same mother and father, fail to produce similar
personalities with any degree of consistency.

The general belief that the behavior patterns of one-egg twins re-
semble each other chiefly because of unusual similarity in their early
environments, both prenatal and postnatal, has yet to be substantiated.
If confirmed by well-controlled twin studies, this knowledge will serve
to strengthen any correctly formulated genetic hypothesis, either concern-
ing normal behavior variations or specific types of mental disorder. In
fact, it is possible that a disordered behavior pattern may result more
immediately from some primary gene effect than can be assumed for a
correlated anatomical defect.

Even if comparative twin studies concentrating on the search for bio-
chemical correlates of basic personality variations have no spectacular
success in the very near future, there can be no doubt about their poten-
tial value for the understanding of human behavior. Important leads
may be obtained by a series of well-planned investigations focused on
psychogenetic features that tend to be relatively constant and provide
evidence of both homogeneity within families and a high concordance
rate in one-egg twins. Most useful, also, will be any study (Allen) that
succeeds in separating genetic and nongenetic components of personality
development by tests for genetic linkage with blood groups and other
easily identified genes.

Obviously, every one of these investigations would require a well-
coordinated interdisciplinary research team, as well as the application of
flexible research techniques lending themselves to optimal use of large

356 FRAXZ J. KALLMAXN

numbers of twin subjects and their families. In planning such projects,
it would be advisable to concentrate adequate facilities and personnel in
a few strategically situated research centers. Instead of encouraging need-
less duplication of studies that may be limited in scope and procurable
support, it would be well to devote all regional resources to the organiza-
tion of a few broad projects., and the training of men qualified to handle
them.

Above all, more than demonstrating that hereditary elements play an
important part in specific behavior variations, the main objective of psy-
chogenetic twin studies should be to demonstrate precisely how this
action takes place.

SCHEMATIC RECAPITULATION

The purpose of this contribution has been to determine the place of
psychogenetic concepts in the over-all theoretical scheme of psychology
as a branch of the human sciences.

The relevance of genetic data in the uses of psychology is based on
the indispensability of this segment of knowledge in the understanding of
ever}7 human function.

It has been shown that twin studies as a research tool are essential
in demonstrating that heredity plays a vital role in potentializing all
basic functions necessary for the establishment of health and the pattern-
ing of normal behavior.

In reviewing the body of information acquired by means of the twin-
study method, the material was organized as follows :

General methodological principles. Twin studies are based on the
regular occurrence of two genetically different types of twins (one-egg
and two-egg). They are applicable in three different versions: (1) the
twin-study method proper, (2) the co-twin-control method, (3) the
twin-family method.

Determination of zygosity. The most reliable criteria in the com-
parative scheme of the modern similarity method are dermatoglyphic
and hematological data. Reciprocal skin grafts can be used if these tests
are indecisive in differentiating same-sex twins.

Ascertainment and analysis of twin samples. Adequate sampling pro-
cedures are important because generalized conclusions should not be
drawn from observations made in single pairs or in an unrepresentative
series of pairs. The statistics describing such a sample are computed from
twin index cases rather than twin pairs. In the analysis of normal per-
sonality traits, twin data are expressed in terms of varying degrees of
intrapair similarity or dissimilarity, rather than in terms of concordance
or discordance (as used in comparing differences in morbidity risks) .

Psycho genetic Studies of Twins 357

Procedural limitations and advantages. The limitations are ( 1 ) Im-
perfections of the research species, ( 2 } imperfections of pliiridisciplinary
research workers, research methods, and research teams, (3) Imperfec-
tions of quantifying methods for measuring meaningful personality dif-
ferences In genetically similar or dissimilar phenotypes.

The advantages are (1) the humanness of the research subjects, (2)
the interdisciplinary nature of the method, (3) its effectiveness as a
sampling procedure and an economical substitute for a total population
survey, (4) the facilitation of combined cross-sectional and longitudinal
studies In a family setting.

Serial twin data on intellectual and personality variations. Two-egg
twins of the same sex tend to differ as much in their personalities and
behavior patterns as any siblings reared together or apart. Consistent
similarity in basic personality traits is found only in one-egg twins and
in them is not erased even by different environments.

Serial twin data on psychopathological variations. Differences be-
tween the two zygosity groups have been found in the concordance rates
of the following conditions: psychoneurotic reaction potentials, male
homosexuality, mongolism, convulsive disease, schizophrenia, manic-
depressive psychosis, and involutional psychosis.

Prospects for future twin research. Twin studies are expected to
make further important contributions to the identification of the nature
and action of genetic components in normal and abnormal behavior
patterns.

In conclusion, it may be said that the techniques of psychogenetics
establish cohesion between two coordinate sciences, psychology and
human genetics. These two sciences belong together and have to pull
together in advancing the understanding of behavioral variations in man.

REFERENCES

1. Address of His Holiness Pope Pius XII. Eugen. News, 1953, 37,
146-149.

2. Allen, G. The meaning of concordance and discordance in estimation
of penetrance and gene frequency. Amer. J. hum. Genet., 1952, 4, 155-172.

3. Allen, G. Comments on the analysis of twin samples. A. Ge. Me. Ge.,
1955, 4, 143-160.

4. Allen, G. Cases of cerebral palsy in a series of mentally defective
twins. Amer. J. ment. Def., 1955, 59, 629-639.

5. Allen, G. Differences between fraternal twins. /. Hered., 1955, 46,
262.

6. Allen, G. Genetic aspects of mental disorder. In Proc. 1956 Annu.
Conf. on The nature and transmission of the genetic and cultural char-

358 FRANZ J. KALLMANN

acteristics of human populations. New York: Milbank Memorial Fund,
1957.

7. Allen, G.? & Baroff, G. S. Mongoloid twins and their siblings. Acta
Genet., 1955, 5, 294-326.

8. Allen, G., & Kaiimann, F. J. Frequency and types of mental re-
tardation. Amer. /. hum. Genet., 1955, 7, 15-20.

9. AnastasL Anne. In D. Hooker (Ed.:, Genetics and the inheritance of
integrated neurological and psychiatric patterns. Vol. 33. Baltimore:
Williams & Wilkins!"l954.

10. Baroff. G. S. A psychomotor, psychometric, and projective study of
mentally defective twins. Unpublished doctoral dissertation. New York
Univer., 1955.

11. Bauer, J. Similarities and differences in the personality and manifest
behavior of one set of monozygotic twins: a depth study involving inter-
acting organic, developmental, psychodynaniic, and environmental factors
as they pertain to the total life history. Unpublished doctoral dissertation,
New York Univer., 1952.

12. Beadle, G. W. Uniqueness of man. Science, 1957, 125, 9-11.

13. Burks, B. S.3 & Roe, A. Studies of identical twins reared apart.
Psychol Monogr., 1949, 63, 1-62.

14. Burlingham, D. Twins. New York: International Universities Press,
1952.

15. Cattell, R. B. Research designs in psychological genetics with special
reference to the multiple variance method. Amer. J. hum. Genet,, 1953, 5?
76-93.

16. Cattell, R. B., et al. The inheritance of personality: a multiple
variance analysis determination of approximate nature-nurture ratios for
primary personality factors Q-data. Amer. J. hum. Genet., 1955, 7, 122-146.

17. Conrad, C. Die erbliche Fallsucht In A. Guett (Ed.), Handbuch
der Erbkrankheiten. Vol. 3, Part 2. Leipzig: G. Thieme, 1940.

18. Cronbach, L. J. Assessment of individual differences. Annu. Rev.
Psychol, 1956, 7, 173-196.

19. Cummins, H., & Midlo, C. Fingerprints., palms and sole. New York:
McGraw-Hill, 1943.

20. Dahlberg, G. Twin births and twins from a hereditary point of view.
Stockholm: Bokforlage-A.-B. Tidens Tryckeri, 1926.

21. Dahlberg, G. tJber genotypische Asymmetrien mit kefalkaudaler
Vererbung. Upsala IdL for en. forh., 1942, 48, 195-220.

22. Darlington, C. D. Heredity and environment. Carylogia, 1954, 6,
370-374.

23. David, H. P., & von Bracken, H. Perspectives in personality theory.
New York: Basic Books, 1957.

24. Dobzhansky, T. Evolution, genetics, and man. New York: Wiley,
1955.

25. Dunn, L. C. The prospects for genetic improvement. Eugen. Quart.,
1956, 3, 32-37.

Psychogenetic Studies of Twins 359

26. Essen-Moller, E. Psychiatrische Untersuchungen an einer Serie von
Zwillingen. A. Psychiat. Neur., Supplement XXIII, 1941.

27. Eysenck, H. J,, & PrelL D. B. The Inheritance of neuroticism: an
experimental study. /. ment. Set., 1951, 97, 441-465.

28. Feingold, L. A psychometric study of senescent twins. Unpublished
doctoral dissertation, Columbia Univer., 1950.

29. Fisher, R. A. Standard calculations for evaluating a blood group
system. Heredity, 1951, 5, 95-102.

30. Ford, N., & Frumkin, S. Monozygosity in mongoloid twins, Amer. J.
Dis. Child., 1941, 63, 847-857.

31. Fuller, J. L. Nature and nurture: a modern synthesis. New York:
Doubleday, 1954.

32. Galton, F. Inquiries into human faculty. New York: Macmillan,
1883.

33. Gedda, L. Studio dei gemelli. Rome: Edlzioni Orizzonte Medico,
1952.

34. Gerard, R. W. Units and concepts of biology. Science, 19573 125,
429-433.

35. Gesell, A. Mental and physical correspondence in twins. ScL Mon.,
1922, 14, 205-244.

36. Gesell, A., & Thompson, H. Twins T and C from infancy to
adolescence: a bio genetic study of individual differences by the method of
co-twin control. Provincetown, Mass.: The Journal Press, 1941.

37. Goldschmidt, R. B. Different philosophies of genetics. Science, 1954,
119, 703-710.

38. Graewe, H. Zwillinge und Schule. Erfurt: K. Stenger, 1938.

39. Grebe, H. Sport bei Zwillingen. A. Ge. Me. Ge., 1955, 4, 275-295.

40. Herrman, L., & Hogben, L. The intellectual resemblance of twins.
Proc. roy. Soc. Edinb., 1933, 53, 105-109.

41. Husen, T. Uber die Begabung von Zwillingen. Psychol. Beitrage,
1954, 1, 137-145.

42. Kallmann, F. J. Twin studies in relation to adjustive problems in
man. Trans. N.Y. Acad. ScL, 1951, 13, 270-275.

43. Kallmann, F. J. Heredity in health and mental disorder. New York:
Norton, 1953.

44. Kallmann., F. J. Twin data in the analysis of mechanisms of in-
heritance. Amer. J. hum. Genet., 1954, 6, 157-162.

45. Kallmann, F. J. The genetics of human behavior. Amer. J. Psychiat..,
1956, 113, 496-501.

46. Kallmann5 F. J. Twin data on the genetics of ageing. Gib a Founda-
tion colloquia on ageing, Vol. 3. 1957.

47. Kallmann,, F. J., & Baroff, G. S. Abnormalities of behavior (in the
light of psychogenetic studies). Annu. Rev. Psychol., 1955, 6, 297-326.

48. Kallmann, F. J., & Rainer, J. Genetics and demography. Symposium
on demography as a science. Univer. of Chicago, Population Research and
Training Center, in press.

360 FRANZ J. KALLMANN

49. Kallmann, F. J,, £ Reisner, D. Twin studies on the significance
of genetic factors in tuberculosis. Amer. Rev. Tuber cuL, 1943, 47, 549-574.

50. Kallxnann, F. J., & Sander, G. Twin studies on senescence. Amer.
J. Psychiat., 1949, 106, 29-36.

51. Kam, M. N. Data of twin births occurring in 2 English metropolitan
hospitals. A. Ge. Me. Ge., 1953, 2, 152-163.

52. Klapper. Z., & Werner, H. Developmental deviations in brain-in-
jured (cerebral palsied) members of pairs of identical twins. Quart. J. Child
Behav., 1950, 2, 288-313.

53. Kranz. H. Lebensschicksale krimineller Zwillinge. Berlin: Springer,
1936.

54. Lamy, M. Les Jumeaux. Paris: Editions Correa, 1944.

55. Landauer, W. The phenotypic modification of hereditary poly-
daciylisrn of fowl by selection and by insulism. Genetics, 1948, 33, 133-157.

56. Lange, J. Verbrechen als SchicksaL Leipzig: G. Thieme, 1929.

57. Lauterbach, C. E. Studies in twin resemblances. Genetics 3 1925,
10, 525-568, 1925.

58. Lennox, W. G., & Jolly, D. H. Seizures, brain waves and intelligence
scores of epileptic twins. In D. Hooker and G. C. Hare (Eds.), Genetics
and the inheritance of integrated neurological and psychiatric patterns.
Baltimore: Williams & Wilkins, 1954.

59. Lennox, W. G., et al. Inheritance of cerebral dysrhythmia and
epilepsy. Arch. Neurol Psychiat., 1940, 44, 1 155-1 183.

60. Lennox, W. G., et al. Twins, brain waves and epilepsy. Arch. Neurol.
Psychiat.} 1942, 47, 702-706.

61. Lilienfeld, A. M., & Pasamanick, B. A study of variations in the
frequency of twin births by race and socio-economic status. Amer. J. hum.
Genet., 1955, 7, 204-217.

62. Luxenburger, H. Psychiatrisch-neurologische Zwillingspathologie.
Zbl ges. Neurol. Psychiat., 1930, 56, 145-180.

63. Luxenburger, H. Untersuchungen an schizophrenen Zwillingen und
ihren Geschwistern zur Pruning der Realitat von Manifestationschwan-
kungen. Z. Neurol, 1935, 154, 351-394.

64. Merriman, C. The intellectual resemblance of twins. PsychoL
Monogr., 1924, 33, 1-58.

65. Money, J. Mind-body dualism and the unity of bodymind. Behav.
Sci, 1956, 1, 212-217.

66. Newman, H. H. Multiple human births. New York: Doubleday,
1940.

67. Newman, H. H., et al. Twins: a study of heredity and environment.
Chicago: Univer. of Chicago Press, 1937.

68. Planansky, K. Heredity in schizophrenia. /. new. ment. Dis..> 1955,
122, 121-142.

69. Race, R. R., & Sanger, R. Blood groups in man. (2d ed.) Spring-
field, III: Charles C Thomas, 1954. Chap. 20.

70. Rogers, B. O., & Allen, G. Intolerance of dizygotic human twins to
reciprocal skin homografts. Science., 1955, 1223 158.

Psycho genetic Studies of Twins 361

71. Rogers, C. R. Persons or science? a philosophical question. Amer.
Psychologist, 1955, 10, 267-278.

72. Rosanoff, A. J., & Orr, I. A study of heredity in insanity in the
light of Mendelian theory*. Amer. J. Insan., 1911, 63, 221-261.

73. Rosanoff, A. J., £ Rosanoff, I. A. A study of mental disorders in
twins. J. juv. Res., 1931, 15, 268-270.

74. Rosanoff, A. J., et al. The etiology of so-called schizophrenic psy-
choses with special reference to their occurrence in twins. Amer. J. Psychiat.,
1934, 91, 247-286.

75. Scott, J. P. The place of observation in biological and psychological
science. Amer. Psychologist, 1955, 10, 61-68.

76. Shields, J. Personality differences and neurotic traits in normal twin
school children: a study in psychiatric genetics. Eugen. Rev.y 1953, 45,
213-246.

77. Siemens, H. W. Zwillingspathologie. Berlin: Springer, 1924.

78. Skinner, B. F. A case history in scientific method. In S. Koch (Ed.),
Psychology: a study of a science. Vol. 2. New York: McGraw-Hill, 1959.

79. Slater, E. Twin research in psychiatry. /. N enrol. Psychiat., 1938,
1, 239-258.

80. Slater, E. An investigation into psychotic and neurotic twins.
London: Univer. of London Press, 1951.

81. Slater, E. Psychotic and neurotic illnesses in twins. Great Britain,
Medical Research Council Special Report Series, 1953, No. 278, Part 2.

82. Slater, E. Psychiatry. In A. Sorsby (Ed.), Clinical genetics. St.
Louis: C. V. Mosby, 1953.

83. Strandskov, H. H. Some aspects of the genetics and evolution of
man's behavioral characteristics. Eugen. Quart., 1955, 2, 152-161.

84. Stromgren, E. Beitrage zur psychiatrischen Erblehre, auf Grund von
Untersuchungen an einer Inselbevolkerung. Ada Psychiat., 1938, Supple-
ment 19.

85. Stumpfl, F. Die Ursprunge des Verbrechens, dargestellt am Lebens-
lauf von Zwillingen. Leipzig: G. Thieme, 1936.

86. Sutton, H. The hereditary abilities study. Set. Rev., 1955, 16, 34-35.

87. Sutton, H., et al. The use of multi-allele genetic characters in the
diagnosis of twin zygosity. Amer. J. hum. Genet., 1955, 7, 180-188.

88. Thorndike, E. L. Measurement of twins. Arch. Phil. Psychol. sci.
Method, 1905, 1, 1-64.

89. Thums, K. Die angeborene zerebrale Kinderlahmung. Allg. Z.
Psychiat., 1939, 112, 262-293.

90. Thurstone, L. L, et al. A psychological study of twins. I. Distribution
of absolute twin differences for identical and fraternal twins. Report No. 4,
Univer. of North Carolina, Psychometric Laboratory, 1953.

91. Vandenberg, S. G. A comparison of identical and fraternal twins on
a battery of psychological tests: a preliminary report on the University of
Michigan twin study. Paper presented at A.I.B.S. meeting, East Lansing,
Mich., 1955.

362 FRANZ J. KALLMANN

92. Von Verschuer, O. Die Ahnlichkeitsdiagnose der Eineiigkeit von
Zwillingen. Anthr. Anz., 1928, 5.

93. Von Verschuer, O. Die Z\villingsforschung als Methode der Genetik
vom Menschen. S. A. S. (Bologna), 1949, 13, 23-102.

94. Von Verschuer, O. Die Erbanlage als bestimmende Kraft auf dem
Lebenswege. Analecta Genetica, 1954, 1, 132-134.

95. Von Verschuer, O. W irksome Faktoren im Leben des Menschen.
Wiesbaden: F. Steiner, 1954.

96. Von Verschuer, O. Erbgebundenheit und Freiheit des Menschen in
seiner EntwickJung. In H. Decher (Ed.), Vom Unbelebten zum Lebendigen,
eine Ringvorlesung. Stuttgart: Ernke, 1956.

97. Weinberg, W. Mathematische Grundlagen der Probandenmethode.
Z. Indukt. Abstamm., 1927; 48, 179-228.

98. Wendt, G. G. Der individuelle Mustenvert der Fingerleisten und
seine Vererbung. A. Ge. Me. Ge., 1955, 4, 330-337.

99. Wiener3 A. S. Heredity of the M-N-S blood types. Amer. J. hum.
Genet., 1952, 4, 37-53.

100. Wilson, J. T.j et al. Current trends in psychology and the be-
havioral sciences. Pittsburgh, Pa.: Univer. of Pittsburgh Press, 1954.

101. Wingfield, A. H. Twins and orphans. London: Dent, 1928.

102. Wright, S. An analysis of variability in number of digits in an in-
bred strain of guinea pigs. Genetics, 1934, 19, 506-536.

103. Zazzo, R. Situation gemellaire et development men tale. /. Psychol.
norm, path., 1952, 45, 208-227.

A, PERSPECTIVE ON SOCIAL PSYCHOLOGY

SOLOMON E. ASCH
Swarthrnore College

Introduction 363

A Question of Perspective 367

The Controversy between Individual and Group Psychology 368

The Data of Social Psychology 374

The Study of Social Influences 379

References 383

INTRODUCTION

The data and problems of social psychology have barely been identi-
fied, and the questions that have been studied in the recent period refer
to only a small part of the field. In these circumstances it would be pre-
mature to propound a formal system of the phenomena in this region.
In any case, I cannot lay claim to the possession of such a system. The
task that faces the student is rather how to proceed during the first stages
of exploration in a difficult and uncultivated territory.

Lack of strict knowledge does not, however, imply the absence of a
theoretical direction. Indeed, the investigator in this area begins with a
strong initial orientation. A long tradition of thinking about human
nature precedes him. He brings to his subject matter the doctrines of
man prevailing in his time and the notions derived from his own ex-
perience. He also draws upon the contributions of general psychology.
These sources of observation and of conceptions, which precede investi-
gation and guide it, constitute a kind of theory, which might be called
presystematic.

Social psychology still works largely with borrowed conceptions that
have not been sharply tested on its own grounds. It has not yet achieved
an independent outlook on its data and problems. In what follows
I propose to examine a few important assumptions and to trace the
effects they have exerted. My particular theme will be the relation
that has prevailed and that should obtain between general and social
psychology.

363

364 SOLOMON E. ASCH

No one will question that social and general psychology have much
to do with each other. But the relation between them is not simple. A
few historical remarks may serve to introduce this discussion.

The main aim of social psychology, which is to further a theory of
human nature, has a long history. But the way in wrhich it proposed to
go about realizing this aim was unprecedented. The new social psychol-
ogy was committed to seek for answers by means of the methods of sci-
ence, of controlled observation, where possible of experimentation, and
to thus bridge the gap between our understanding of natural and social
events. The idea of such a discipline wras the culmination of a series of
great changes in thinking about nature and society.

At the same time, this movement was in part a reaction against the
narrowness of a general psychology which found no place in its scheme
for some of the most essential properties of men. The scientific psychology
from which it sprang restricted its observations to the relations between
an individual and an environment that strictly excluded other persons;
it was not concerned with relations between persons or between persons
and groups. The movement toward a social psychology represented an
insistence that these major and neglected parts of human psychology be
taken seriously. It stood for the belief that no psychology can be complete
that fails to look directly at man as a social being.

There were a number of reasons for this restriction in the scope of
psychology during the first fifty years of its existence. Investigators had
no techniques for the exact investigation of psychosocial phenomena ; and
like most humans, they preferred safe and tested procedures. Probably
they also feared the complexity of social events, and saw little hope of
studying them in the manner that their conception of science demanded.
The belief that this area was outside the range of experimental analysis
discouraged further interest.1

It is noteworthy that the new social psychology did not start with a
commanding discovery which could guide thinking and investigation, and
furnish an answer to the preceding doubts. It was rather the expression
of a hope that the procedures of observation and experimentation were
not limited to selected phenomena within our field, and of a determina-
tion to demonstrate this faith in practice.

Let us now consider the ways in which the relations between general
and social psychology have been conceived. Since there has been little
explicit discussion of this question, it will be necessary to infer the rele-
vant views from existing trends of investigation and theory. At this point
one finds quite diverse emphases.

1. There are those, both within and outside social psychology, who

1 Consider, for example, the position that Wundt espoused in the Volker-
psychologie.

A Perspective on Social Psychology 365

hold that it is an applied discipline, that it, involves little more than the
application of some of the laws of "nonsocial" psychology to more com-
plex data. There are two main grounds for this position. First, it is clear
that the same principles of human functioning are equally at work in
the social and nonsocial settings. The principles of learning or of motiva-
tion discovered by the procedures of strictly individual investigation must
also be valid under social conditions. Few, if any, will dispute this
thought. The other ground of this position is less often stated openly.
This is the belief that all principles of psychological functioning will be
discovered in the nonsocial setting. This belief follows from the further
assumption that the basic data of psychology are those that concern the
most elementary phenomena, and that more extended phenomena are
complications of these. This was the view of Hull [5], w^ho held that
social and moral data can be derived from a knowledge of basic learn-
ing principles. It is also the position that Skinner [10] maintains. It
seems to follow that social psychology (and perhaps also the other social
disciplines) has no basic theoretical problems of its own. In apparent
support of this belief is the further observation that the term "social" does
not designate a particular psychological function, such as memory or
perception, but rather includes all. On these grounds the following
division of labor is recommended: Let general psychology discover the
principles, and let social psychology extrapolate them. This has continued
to be a leading emphasis within social psychology itself.

There is much to be done in the way of relating what is known in
general psychology to social data. But the second assumption simply
denies, with not even an attempt at proof, that the urgent problem is to
advance our meager understanding of the complex cognitive and
emotional operations on which social events rest. An enterprise that starts
on such a precarious footing is not likely to rise above its source, or move
toward a coherent body of knowledge. Furthermore, will not those who
are attracted to psychology prefer the challenge of building the founda-
tions to the work of premature and dubious application?

2. There is also a more pragmatic trend in social psychology. In
every area of study there is much to be done in the way of observation
and description, and of establishing particular formulations, without
the necessity of facing immediately their relations to more fundamental
propositions, in the present instance to those of general psychology.
From the standpoint of practice there is much to be said for this direct
approach. Work done under such auspices can be useful; surely we can
profit from increased knowledge of shifts in public opinion, of interrela-
tions among opinions., of the distribution of prejudice, of relations be-
tween leaders and followers. In the long range, however, it reveals a
weakness. It assumes that the study of social-psychological events poses

366 SOLOMON E. ASCH

no problems of basic clarification, that it is enough, to have innumerable
questions one can ask about social behavior and experience and to possess
rules of method that secure the objectivity of one's findings. The danger
of this position is that it creates a technology7 before there is a science.

3. Another view, and one that I favor, may be stated as follows.
Social psychology is not an applied discipline. Its task is to contribute to
a theory of the psychological functions. This cannot be accomplished
only by studying individuals in exclusively individual settings; it requires
also the direct investigation of happenings between persons, or the ex-
tension of observation beyond the limits that experimental psychology
had traditionally imposed. We cannot have a tenable theory of emotions
or motives if we do not study those that refer directly to persons; no
procedure of extrapolation will suffice for this purpose. It would be
most unusual if we had discovered the key to the central properties of
men without having to go to the trouble of studying them in the only
setting in which they can be observed. I hold that social psychology is
part and parcel of the enterprise of general psychology.

It follows that social psychology is under obligation to make its own
contribution to the persistent problems of general psychology. This state-
ment surely does not describe what has happened during recent decades,
and may appear to be an expression of an unrealistic hope. Let us, how-
ever, push the conclusion to its limits. The paradox of this conclusion is
that it demands an independent, critical examination of the basic issues
of psychology in the light of the data of social behavior and experience.
It asserts that if the need for a social psychology can be traced to the
lack of an adequate base in general psychology, the gap cannot be
remedied by a wholly derivative discipline.

Much that has happened in this field is, I would say, the result of
a failure to take this possibility seriously. Although social psychology was
partly a revolt against the existing order, it nevertheless grew up in the
shadow of general psychology, from which it borrowed its concepts and
procedures. It has adopted existing formulations about the operations of
motives, emotions, thinking, and learning, which were derived mainly from
the study of strictly individual, indeed non-human, behavior, and has
systematically confined its investigations within the prevailing frame of
concepts. It will be my contention that this dependence has been responsible
for the neglect of some central questions and for a limited horizon.

These comments follow from my estimation of the achievements of
social psychology, especially during the recent decades which were
marked by an unusual expansion of interest and activity. I come away
with two distinct impressions. First, one must record a number of gains.
There has been a sharpening of problems, a growth of techniques; and
some additions to the body of knowledge have been secured. An opti-

A Perspective on Social Psychology 367

mistic assessment of the situation is therefore not difficult to reach. One
may admit that the advances have been limited, but this is not unusual
at certain stages of a science. Above all, many will be inclined to stress
the power inherent in empirical procedures to replace vague generalities
with tested knowledge. "Better a minute truth than a grand half-truth"
expresses fairly well the prevailing spirit. The second impression goes in a
quite different, indeed disturbing, direction. There is something puzzling
about today's social psychology. Much careful and conscientious work is
going on, but it is hard to avoid the conclusion that it has not been
fruitful in fundamental conceptions. It has not produced many instances
of decisive research, and has not perceptibly deepened our knowledge
of man, despite the very considerable effort that has undeniably been
expended.

In what follows I have chosen to discuss in this light the adequacy
of the steps we have taken in studying a few problems that have been
our chief concern.

A QUESTION OF PERSPECTIVE

Let me begin with a very general point. Each discipline possesses its
special spirit, which consists in a particular way of viewing its data. The
study of man as a social being also requires its own perspective, which
must start from some conception, however tentative, of what it is to be
human. The subject who sits for our portrait, Homo sapiens, is, to be
sure, only dimly visible to us, and we will probably not produce a good
likeness of him soon. But even a first sketch requires some apprehension
of his dimensions. These would have to include as a minimum that he
possesses unusual intellectual powers, that he can act with reference to
ideas and ideals of right and wrong, even when he violates them; that he
has a need to surround himself with objects that are attractive. To
realize that these are part of the "human minimum" is essential as a
point of departure for thinking about man.

Today there seems to be little evidence of this awareness. The
question, what it is to be human, which should be of more than passing
concern, has virtually disappeared from discussion. At the same time,
the man of social psychology turns out to be a quite dwarflike creature.
One would not often suspect that we were talking of an organism capable
of keeping or betraying faith with others, in whose history religious
beliefs have played quite a part, who can cry out for justice. It is hard
to escape the conclusion that primitive notions of what it is to be
human have guided thinking and investigation.

It is not my intention to say that social psychology should have
prompt answers to the most difficult questions. I am suggesting that a

368 SOLOMON E. ASCH

certain broadmlndedness about human ways is necessary for thinking
of the required scope. We cannot be true to a fragment of man if we
are not trae5 at least in a rudimentary way, to man himself. The con-
sequence of this neglect is aimlessness in investigation and fragmenta-
tion of knowledge. The danger of ignoring relevant matters is that less
relevant matters ucurp their place. Actually many of us today do have
a point of departure which follows the curious doctrine that man is
directly descended from the white rat. A limited perspective can have the
effect of trivializing a subject. Those who deepen our understanding help
us, as a rule, to see more in a given region than we had suspected to be
there; in social psychology one often has the uncomfortable feeling of
being invited to see less than we thought was before us. One wonders
whether a failure to appreciate the relevance of history and literature,
of art and religion, to human affairs is the prerequisite for the pursuit of
social psychology.

We will now abandon general statements to consider how social
psychology has dealt with several particular problems, and the role that
general psychological theory has played. I have chosen three problems
for illustration, although others might have served equally well.

THE CONTROVERSY BETWEEN INDIVIDUAL
AND GROUP PSYCHOLOGY

Let us consider the fate of a problem in motivation which belongs
at the very center of the discipline. It is of considerable consequence for
any social psychology to establish the grounds of concern for the welfare
of other persons or groups, and how these are related to the concern
individuals feel for their own welfare. What has happened to this prob-
lem? It has virtually disappeared or been interpreted out of existence, one
suspects for no better reason than that it has not attracted the interest
of general psychology, which could offer little guidance at this point just
because it had excluded the relevant phenomena from view.

It may be well to trace the course of thinking about this question
historically, since it is connected with an important set of problems that
came to a head when the first steps toward a social psychology were
taken. It is not my intention to examine the evidence, but rather to
ask what circumstances decided the way in which social psychology went
about defining its task.

Our story begins with the controversy centering around the con-
flicting claims of individual psychology and of group psychology. The
issues entering into this well-known discussion have not, I believe, been
fully understood. This is mainly because some of the best-known pro-
ponents of a group psychology were driven to the expedient of postulating

A Perspective on Social Psychology 369

a group mind which, as every proper social psychologist today has
learned, is a mystical notion. Actually this was a relatively superficial
feature of the dispute; lying just underneath the surface were the serious
issues, still very much alive today.

Why was the group mind thesis put forward by able men? It started
with a serious problem — with the clarification of group characteristics
and group membership. It had its roots in a formulation by no means
strange today, namely, that one cannot understand an individual by
studying him solely as an individual; one must see him in his group
relations. Thinkers like McDougall were especially impressed by the
phenomenon of group spirit. Members of organized groups, they held,
were guided in their sentiments and actions by the idea of the group.
That is to say, when acting as group members, they were orienting
themselves to a reality vastly more powerful than themselves, and one
that was able to command their devotion and interest. Since McDougall
and others saw no way of deriving these great social forces from the
properties of individuals taken separately, or from the aggregation of
individual characteristics, they felt compelled to postulate a group
mind.

From within psychology there came a sharp response to the group
mind doctrine. It was clear that there can be no psychology other than
that of the individual. What, then, is social psychology? The answer
was stated most clearly in the 1920s by F. H. Allport.

Allport went far beyond the rejection of a group mind and the as-
sertion that social psychology is the study of individual behavior (and
experience). This first step only cleared the way for the second, and far
more important, formulation — that concerning the nature of individual
behavior and experience. There is, to begin with, the following account
of

. . . the essential formula for behavior: ... (1) Some need is present in
the organism, such as the necessity of withdrawing from weapons injuring
the body, or the need to obtain food, or to secure a mate. (2) The organism
acts: it behaves in such a manner as to satisfy the need [1, p. 1].

This statement cleared the way for the characterization of the sub-
ject matter of social psychology:

Social behavior comprises the stimulations and reactions arising between
an individual and ... his fellows. . . . The significance of social be-
havior is exactly the same as that of nonsocial, namely, the correction of
the individual's biological maladjustment to .his environment. ... In and
through others many of our most urgent wants are fulfilled; and our be-
havior toward them is based on the same fundamental needs as our re-

370 SOLOMON E. ASGH

actions toward all objects, social or non-social It is the satisfaction of these
needs and the adaptation of the individual to his whole environment which
constitute the guiding principles of his interactions with Ms fellow men
[1, pp. 3-4].

It was this formulation (representative of the then dominant general
psychology) that won the day. If one abstracts from details, it still ex-
presses the spirit and embodies the assumptions of virtually all con-
temporary social psychology. For this reason alone it deserves the closest
examination.

The main point may be restated as follows: all that a person does,
all that he feels and thinks is determined by the tendency to gain satis-
faction for Ms needs. AUport was thus asserting a basic proposition about
human motives: the egocentric proposition. It is here, in this char-
acterization of what an "individual" is, that we must see the root of All-
port's disagreement with group mind theorists. The latter were seeking a
way to express what they sensed to be an essential feature of social life —
the capacity of individuals under- some circumstances to transcend their
own particular interests and to act in the interest of their group. The
current general psychology saw this as a false problem for two closely
connected reasons. It denied the reality of groups on elementaristic
grounds, and could only conclude that the idea of the group was an
illusion of individuals. But, in addition, it defined at the outset an all-
inclusive property of human motives in such a way as to require a re-
interpretation of all group sentiments in terms of self-centered motives.

We can see that there was indeed a big difference between this psy-
chology of the individual and the group psychology that some thinkers
considered necessary for a complete account of human behavior.

This psychology of the individual also defined in a particular way
the scope of social psychology. The latter was to be one small comer
of psychology which, instead of studying the usual stimuli — weights,
lights, sounds — dealt with social stimuli. The other person, too, is a
stimulus. But if "the individual in the crowd behaves just as he would
behave alone, only more so" [1, p. 295] ; if it was the "individual citizen35
who stormed the Bastille, one could only conclude that social phenomena
were not of major theoretical interest.

As often happens in the history of thought, conflicting doctrines may
be at one in the most important assumptions. This was the case with the
group mind and the individualistic formulations. The problem they were
facing was that of reconciling an apparent antinomy: that men are social
beings and that they lead an ultimately private existence. Both failed
to describe the process that overcomes this paradox and that is responsible
for group phenomena at the psychological level Social action requires

A Perspective on Social Psychology 371

that the individual participant be capable of representing to himself the
situation that includes himself and others. These individual representa-
tions contain, in cases of full-fledged interaction, a reference to the fact
that the others also possess a corresponding view of the situation. These
similar and mutually relevant representations in individuals provide the
equivalent of what group mind theorists sought and individual psy-
chologists denied, [See 2, chaps. 5, 10, 1 1 for a fuller account]

In short, social action in humans rests on an unusual kind of part-
whole relation, in which the structure of group conditions is (at least in
part) represented in the individual, who can only in this way become
a participant. Consequently a group does not need a central head-
quarters, and a social act is not reducible to the model of action between
person and thing. Neither of the contending doctrines saw this solution.

In the recent period, some social psychologists have come to adopt
the view that psychosocial events are based on such similar and mutually
relevant representations in individuals. They have accepted it mainly,
however, with reference to the cognitive side of our functioning, while
retaining intact an exclusively egocentric conception of motives. This
procedure fails to draw the full consequences of the formulation. If the
representations by individuals of their relation to others are to issue in
action that is not chaotic, they must have mutual reference in some de-
gree. Is this possible if each person as a rule sees a given situation solely
from the standpoint of his needs? The hypothesis we are forced to con-
sider is that mutually relevant fields are not consistent either with a
purely egocentric account of cognitive or of motivational events.

An error in thinking and in psychological analysis made it appear
that there is no alternative to the egocentric formulation. To be sure,
mothers have been known to starve in order to feed their children, and
persons have endangered their lives for others. There are, then, actions
which at least appear to be quite the opposite of self-centered. But con-
trary data are weak reeds against winds of doctrine. The ready answer
stood at hand that the need to help others is egotistic because one enjoys
it; it is egotistic to enjoy one's unegotistic action. The error of this too
clever argument lies in the refusal to face and explore an intelligible alter-
native, in the failure to admit as a legitimate possibility that under certain
conditions the place of egocentric needs in the individual's brain is not
functionally at the center, but that egocentric needs may themselves be
localized in the brain as dependent parts of a wider situation.

The issue is, of course, a factual one; conceivably the most seemingly
disinterested action may be the work of calculation and self-interest.
Admittedly a decision about this question is difficult, perhaps mainly be-
cause much of human behavior is a function of both kinds of vectors.

372 SOLOMON E. ASGH

But science does not justify dogmatism when a problem is beset with
difficulties.

The point I wish to make is that social psychology has passively
accepted a prevailing view, of the greatest consequence to its subject
matter, without taking the trouble to consider observations and distinc-
tions which should be the starting point for thinking. It ignored observa-
tion in favor of a theory that replaced observation. It is true that if the
secret police knock on a door in the dead of night, the neighbors may be
alarmed for themselves. Is it equally clear that they have no concern for
the victim? Where is the stringent evidence that a sense of injustice at the
mistreatment of another plays no role at all?2

An error about a fundamental proposition such as we have been
discussing, one wrhich concerns the possible kinds of relations between
persons, is bound to have consequences. If group events require that
persons should feel and act as part of their group, and not solely as the
center of happenings, then the error is of the same magnitude as that
of a doctrine that would deny the self-centered tendencies of persons and
describe society as an exercise in altruism.

This omission has been responsible for a limited and lackluster treat-
ment of central topics. To it we should trace the accounts of group be-
longing as a kind of business transaction obeying the motivational
principles of a watered-down Hobbes; one hardly finds an inkling in
these studies that being in a group can be either an enjoyable or a
responsible experience. It must be credited also with the formulations
about attitudes that give the lion's share to their opportunism, and with
the neglect of those attitudinal forces that take possession of the person,
including the part we call his self. It has dampened concern with those
values that appear to be determined chiefly by objective requirements.

A reexamination of a range of problems in the light of this issue
could prove a challenging task. One may safely say that if social psy-
chology is to make progress, it must take into account the vectors that
make it possible for persons to think and care and work for others. It
will need to find a place for the capacity of persons to relate to the needs
of a situation so that they become the needs of the person; it will have
to acknowledge that the desire to play one's part meaningfully may at

2 It might avoid misunderstanding of what has been said to add that I have
not questioned the power of egocentric needs. The preceding discussion might
become more palatable if it were pointed out that action determined by the
interests of others is not necessarily wise, nor are its consequences unfailingly bene-
ficial. Indeed, actions in the interests of one's immediate group may be aggressive
and destructive to outsiders. With these remarks I may perhaps be exonerated
from defending a "soft" doctrine.

.4 Perspective on Social Psychology 373

times be strong, and that It may even make sense to speak of an in-
dividual's desire for the realization of a better society.

Before leaving this topic two further remarks may be In order.

The first concerns a reformulation of the content of mutually relevant
fields in the light of this discussion. One dimension of the consensus they
produce has to do with the intellectual assessment of a given situation :
There can be no concerted action between persons unless they have
cognitively structured the given conditions in somewhat similar ways.
( Included in the given conditions are the actions and intentions of the
participants themselves. There must be a similar way of understanding
both the material properties of the environment and the psychological
properties of the participants.) But social action cannot get started on
the basis of an intellectual appreciation of such data alone; there must
also be a degree of affective consensus with respect to the aims and
needs of the participants. The traditional view describes this second
aspect of consensus as a concurrence of ego-centered orientations. The
alternative here discussed is that the need or goal of one person can,
given certain conditions, arouse forces in another person toward fulfilling
them, without exclusive reference to the latter's "own" needs. This rela-
tion to another, when it is mutual and known to be such, seems to me to
be an indispensable condition of mutual trust and of group coherence.

The next remark concerns the relation between the mutually relevant
fields of individuals and a total group process. It follows from what has
been said that group events lack the solid monolithic structure which
they phenomenally give us. A group is not a single physical system; it
does not possess the kind of unity that belongs to a thing or an individual.
For example, the "body of medical knowledge" does not have a single
locus; it is distributed among many individuals and includes what is
to be found in libraries and hospitals. Phenomenally we objectify group
events to a high degree; it would almost be right to say that there are
group minds, but that they exist in individuals, and that there are as
many group minds as individuals in a group. Also, a group event in-
cludes more than the psychological activities of its members. In addition
to the environment — natural and social — and the activities of its par-
ticipants, it includes the structure of initiated events and the regularities
these exhibit, whether or not they are known to the participants. The
study of these regularities is the problem of other social disciplines, such
as anthropology and sociology.

At this point it seems best to stress the distinction between such
a total group process and the psychological components of it. Recently,
Sears has suggested that psychologists who study social behavior and
personality may be at fault in limiting their view to a single individual,

374 SOLOMON E. ASCH

and proposed that we take the dyad as the minimum unit of action. "A
dyadic unit/' he says, "is essential if there is to be any conceptualization
of the relationships between people . . . 33 [9, p. 478]. One can only
welcome an effort to repair the individualism of an earlier period, and to
arrive at an equivalent of group realities in a behavioristic way. At this
point it is necessary to enter a reservation in favor of a more individual
and more complex approach. Here one should refer to three dyadic
structures. One is the inclusive sociological formation; it is right to say
that we must keep it in view if we are to follow the course of action
which individuate jointly produce. But it is necessary to distinguish be-
tween this inclusive structure and the contributions to it by each of the
participants. At this time., I know no way of describing the psychological
and sociological happenings within a single conceptual formulation.

THE DATA OF SOCIAL PSYCHOLOGY

The place of experience in human social psychology has been settled
in a purely practical way. It is not possible, as a rule, to conduct in-
vestigation in social psychology without including a reference to the
experiences of persons. The investigator must, for example, take into ac-
count what the person under observation is saying; and such utterances
have to be treated in terms of their meaning, not as auditory waves, or
sounds, or "verbal behavior." One can hardly take a step in this region
without involving the subject's ideas, feelings, and intentions. We do this
when we observe people exchanging gifts, engaging in an economic
transaction, being hurt by criticism, or taking part in a ritual. The sense
of these actions would disappear the moment we subtracted from our
description the presumed mental operations that they imply. This re-
quirement to include mental happenings in an account of human
activities, one which the social disciplines generally must observe, should
have spurred an examination of the systematic properties of experience
and their relations to action. Instead we find that the situation has been
accepted half-heartedly, and that its implications have not been explored
with care.

To see how the problem of experience arises in this area, let us con-
sider how we follow the actions of persons. The first observation we
make is that persons invariably describe the doings of others (and their
own doings) in psychological terms. We say that a person sees, hears,
prefers, demands. This is also the way we describe happenings between
pereons; thus we say that one person helped another, or distrusted him.
These are the ways in which we order the actions of persons whom we
observe to be living and conscious.

An organism that relates itself to the environment in this manner is

.4 Perspective on Social Psychology 375

observed to act in it in a special way. Heider [4] has pointed out that we
observe persons to produce effects intentionally. They relate themselves
to the environment by wanting, by being interested, by liking, by under-
standing. In the case of persons, a cause is not merely a preceding state of
affairs; it is a state of affairs as known or understood by the actor. An
effect is not merely a later state of affairs; persons make things happen,
or intend them. The movements of persons thus gain the status of actions.

With these is connected the most significant property of persons:
that we experience them as capable of responding to us. They alone can
understand our thoughts and feel our needs. Therefore they become the
adequate objects of praise and blame. It is only to beings having these
properties that we can relate ourselves by cooperation and competition,
by affection and hatred, by admiration and envy. It is in these terms
that we follow the actions of a friend, the happenings in a play of
Sophocles or in the life of a primitive society. Events of this kind form
much of the content of the mutually relevant fields of persons discussed
earlier.

From the standpoint of a powerful tradition there is something sus-
pect about these everyday observations. The main charge is that they
do not speak the language of science. They refer, of course, to what the
other person does, but they are not simply descriptions of the movements
he carries out; they are not simply statements of the geometrical dis-
placements of persons and things. At this point the temper of one theory
in general psychology recommends the wholesale dismissal of the lay-
man's concepts and language when we turn to investigation. His ac-
counts are, it is said, contaminated by the inclusion of subjective condi-
tions that are not observable because they are not describable in terms of
physical operations. This formulation, although it has not originated in
social psychology and would, if taken seriously, drastically curb further
inquiry in this field, has nevertheless left a strong impress upon it.

The following illustration may clarify the point at issue and the dif-
ficulties it raises. Among his prescriptions for a psychological Utopia,
Skinner includes the training of children to tolerate frustration, and
proposes an ingenious procedure [10]. He would occasionally have the
children in his Utopia come to their meals, but delay their eating for a
few minutes while they watch some delicious specialties that had been
prepared for them. Gradually he would extend the period of deprivation,
the intention being to instill self-control without injurious consequences.
As Skinner describes the procedure, it is exclusively an affair of timing
responses to given physical conditions. One may be permitted to wonder
whether the children, however carefully reared, might not take a dif-
ferent view of the proceedings. Are they not likely to wonder what their
caretakers are up to? And will not the outcome depend on the answers

376 SOLOMON E. ASCH

the caretakers give? If it should come into the children's heads that
the caretakers are malicious, it might go ill with the effects of the
scheduling. On the other hand, if the children trusted their mentors, and
particularly, if they credited the caretakers with the meritorious motive
of instilling self-control without injurious consequences, the discipline
might prove more successful. The elimination of any reference to these
internal events amounts to a failure to describe the relevant conditions
with any adequacy.

The problem arises whenever we refer to action between persons. A
determined effort to treat the relation of frustration to aggression in non-
experiential terms could not avoid defining frustration as damage
attributed to a particular instigator [3]. In a recent discussion, while
again insisting that we give priority in psychological investigation to
action, on the ground that it alone is public. Sears uses the following ex-
ample: "... if a child wants to be kissed good-night, his mother must
lean toward him affectionately and kiss him. He, in turn, must slip his
arms around her neck and lift his face to her receptively" [9, p. 480].
This sentence is surely not an unadulterated description of geometrical
displacements; it does not supply the kinematics of affection, or even of
slipping, lifting, or leaning.

The sources of disagreement about the place of experience in psy-
chological investigation are too deep-seated to be dealt with summarily.
We will consider only a few points most relevant to this discussion. In
the first place, there are certain misconceptions to be noted. It is often
asserted that actions are public but experiences are private, and that
therefore the latter have no place in science. Surely there is an error here.
The observation of actions is part of the observer's experience. Indeed,
the same writers who make the first assertion as a rule subscribe to the
second. There is thus no ground for calling actions objective and ex-
perience in general subjective. This confusion has been discussed by
Kohler [7] ; it is not necessary to repeat it in full.

There is another, seemingly more substantial reason for the difference
in status accorded to behavior and experience. We can, it is asserted,
arrive at a high degree of consensus about behavior, but not about our
respective experiences. (In the light of the preceding point, this assertion
claims that some kinds of experience produce consensus superior to
others.) In particular, the conclusion is drawn that the positions and
displacements of objects in space provide the only dependable consensus.

This conclusion will not withstand scrutiny. There is often, indeed,
excellent consensus about events which, according to the preceding view,
are unobservable or incommunicable. The size of an afterimage, or the
experience of a causal connection, can be described with a lawfulness
that permits the study of their dependence on inner and outer conditions.

A Perspective on Social Psychology 377

This suffices to qualify the data of experience as data of science. Instead
of pursuing this fruitful direction, the physicalistic doctrine has attempted
to demonstrate that the data of experience can be treated as verbal be-
havior. It can be shown, though, that the occurrence of an experience is
not the occurrence of a verbal response.

It is not probable that the preceding formal arguments are actually
responsible for the efforts to eliminate all reference to experience from
human investigation. To locate the sources of difficulty we must look else-
where. Perhaps the most decisive assumption is that the data of experi-
ence are not functionally connected with, and provide no help towrard
understanding, other concurrent events in the individual. This belief
is contrary to what we know about the relations of mental and physical
events. The physicalistic program also derives from the elementaristic as-
sumption that the properties of action can be exhaustively described in
terms of component movements. Were this the case, it might indeed fol-
low that the data of experience have a limited place. But human actions
are extended spatio-temporal events having a definite form, and we can-
not describe them without reference to goals, and to means related to
goals. These characteristics of actions are lost from view when we concen-
trate on their most minute components one at a time, just as we lose the
quality of a form or a melody when we attend only to its smallest com-
ponents. It has been convincingly shown that the most consistently be-
havioristic procedures do not actually deal with stimulus and response in
these elementaristic terms [6]. Behaviorism must and does include action;
it grants in practice all that is needed when it speaks of "running toward
a goal," or of "pushing" and "pulling."

What is the relation of the distinction we have tried to draw between
movement and extended action to the data of experience? First, the data
of experience point to, and thus help identify, the conditions in the en-
vironment to which we are responsive. Second, the data of experience
provide hints concerning the internal events that steer action.

Those who dream of an objectivistic social psychology fail to realize
that such a program can be pursued only if the data of experience are
taken into account openly. We are today far from able to describe the
most obvious and the most significant social acts except in the language
of direct experience. What are the event-sequences corresponding to such
data as "the mother praised the child/3 or "the boy refused to heed the
teacher"? And how much more difficult is it to describe the actions of
"keeping a promise" or "telling the truth"? Not only are we at a loss to
report adequately the actual sequences of such events; there is often no
fixed set of actions corresponding to them from occasion to occasion.
How, then, could we go about locating and identifying the relevant
action patterns unless we were guided to them by the distinctions of

378 SOLOMON E. ASCH

direct experience? Even if we succeeded in such a description, it would
remain a foreign language until it was translated back into the terms we
ordinarily employ. At this point the categories of the layman are actually
in advance of those that formal psychology today has at its command.
He has, without the benefit of a psychological education, identified some
of the conditions and consequences of action. To be sure, these categories
are descriptive, not explanatory. Also, everyday thinking identifies them
in a shorthand, summary manner, which must be replaced with far more
detailed description. But to counsel their abandonment is to give up the
prospect of social understanding, and to bar the very advance toward
which we aim.

Throughout this discussion we have noted the prevalence of the
assumption that one can move directly from a few selected notions, de-
rived mainly from the study of lower organisms, to an account of human
actions, and that the latter require no concepts appropriate to them.
Actually, concepts such as conditioning, stimulus generalization, extinc-
tion, response strength, secondary reinforcement, and reinforcement itself
have as a rule been extrapolated to social settings without a serious effort
to demonstrate their relevance under the new conditions. In this passage
the terms lose the relatively clear sense they initially have. The extra-
polations become largely verbal ; we are not the wiser when the transla-
tion has been accomplished. This procedure, instead of increasing ob-
jectivity, often conceals distinctions long familiar to ordinary observa-
tion. It discourages the exploration of those differences between persons
and things, between living and dead, that are at the center of the sub-
ject. It creates the curious presumption that hardly anything new re-
mains to be discovered in a field that has barely been studied.

The conclusion we have reached could have been arrived at more
simply. Every field of inquiry must begin with the phenomena that every-
day experience reveals, and with the distinctions it contains. Further
inquiry may modify our understanding of them, but the phenomena
themselves will never be displaced. In social psychology the phenomena
with which we begin are qualitatively diverse and the description of them
prior to formal investigation is consequently of particular importance.
Let us, for the purpose of this discussion, assume that concepts such as
"role," or "internalizing of values," have a place in social psychology.
They must then be shown to apply to the ways in which the actors, who
are often innocent of these notions, see their situation. The latter act in
terms of conceptions and emotions peculiar to them — in terms of envy
and trust, hope and suspicion. The concepts must be relevant to this
world of appearances, which are among the indispensable data of the
field. Those who avoid this initial phase of investigation run the danger

.4 Perspective on Social Psychology 379

of placing themselves In the position of the hero in Greek mythology
who was shorn of his power the moment he lost contact with mother
earth.

Having said this, it is necessary to add that a psychology based on
phenomenal data alone must remain incomplete. The latter are always
part of a wider field of events within the individual; any order they may
reveal will be partial unless completed by a more comprehensive knowl-
edge of psychological functioning. We need, therefore, an objective psy-
chology that will account for the structure of experience. It also follows
that the examination of experience should not become either an aimless
or an endless occupation. It should strive to issue in inductive inquiry
and, where possible, experimentation.

These conclusions should not hide the difficulties that face investiga-
tion in social psychology. In one area of psychology, that of perception,
the reliance on phenomenal data has proceeded fruitfully. Such investi-
gation possesses one indisputable advantage: phenomenal events are
studied in their dependence on stimulus conditions which are describable
in terms of well-understood physical operations, and in relation to
internal processes that are also described in terms of natural science
categories. This advantage deserts us in most parts of social psychology.
Here we must abandon, at least for the foreseeable future, the yard-
sticks of physics, and describe both the stimulus conditions and the effects
they produce in psychological terms. Since the dimensions of these events
are frequently complex and only vaguely known, the prospect of dis-
covering clear functional relations may arouse skepticism. It would be
misleading to minimize the difficulties, but it would also be premature to
prejudge the outcome. This is a challenge social psychology must accept.

THE STUDY OF SOCIAL INFLUENCES

The final problem in illustration of my theme concerns the study
of social influences. Social psychology recognized early the importance of
this area and has energetically investigated it.

My comments at this point follow directly from what has preceded.
Social influences differ from other conditions — such as heat and cold,
light and dark — mainly in this respect: they are experienced to have
their source in persons. They refer to the purposes, attitudes, and
thoughts of others. It follows that we cannot talk sensibly about the
effects of social conditions without specifying their cognitive and emo-
tional content. One responds differently to the same action, depending
on whether it is judged to be friendly or unfriendly, deliberate or acci-
dental, serious or frivolous. The first consequence of this observation is

380 SOLOMON E. ASGH

to underscore the fact that a nonpsychological definition of the stimulus
conditions and their effects may leave the essentials out of account.

In an effort to bring order into this area investigators have leaned
toward a simple and seemingly comprehensive conception of social in-
fluences. It has two main properties. First, it assimilates all group in-
fluences to the construct of constraint or pressure, and all their conse-
quences to the construct of conformity. Second, it applies a general
proposition about the operation of rewards and punishments to account
for the observed effects.

There is good reason to be skeptical of the assumption that all in-
stances of social determination are of the same kind, differing only in
detail. In particular, it is a serious error to equate social determination
generally with constraint or pressure. I am more impressed by the need
to discriminate the kinds of social influence, and will attempt a few re-
marks concerning this point.

1. One of the great effects of social experience is to produce con-
sensus about considerable reaches of the environment. We discover in
the course of action that many of the reports of others are validated in
our experience, and conversely. It is on this foundation that action be-
tween persons occurs. This form of consensus is largely the product of
observation and judgment. By no stretch of the imagination can it be
equated to the operations of constraint and conformity.

It is a curious fact that contemporary social psychology, which is not
inclined to underestimate the effects of social conditions, has seen no
problem here. The assumption has been that the individual's unaided
experiences suffice to validate the basic features of the physical environ-
ment, and that social effects enter only at the point where he must take
the reports of others on trust. This assumption (which leads necessarily
to a subjectivistic account of social determination) ignores what may be
the crucial point, namely, that we accept the reports of others in lieu of
direct experience only because we have at other times received the most
direct proof of the validity of their reports, that only on this basis do we
extend the area of consensus into what is not directly perceptible.

2. There is another, and quite different, range of operations to
which it makes no sense to apply the notions of constraint and conform-
ity. The actions of persons exert emotional effects upon us. I do not know
what it would mean to say that one is afraid or envious out of social
pressure.

3. Each social order confronts its members with a selected portion
of physical and social data. The most decisive feature of this selectivity is
that it presents conditions lacking in perceptible alternatives. There is no
alternative to the language of one's group, to the kinship relations it
practices, to the diet that nourishes it, to the arts it supports. The field

A Perspective on Social Psychology 381

of the Individual is, especially in a relatively closed society, in large meas-
ure circumscribed by what is included in the given cultural setting.

These conditions produce a kind of socially generated reality, which
Is as much part of the environment as topography and climate. They of
necessity shape the individual's expectations, needs, and character, often
perhaps irreversibly. The consequences are more fundamental than those
generally dealt with by empirical investigation which has been con-
cerned mostly with modifications of already formed views and needs.
From the psychological standpoint the significant feature of these condi-
tions is their monopolistic character, or the absence of known alternatives.
The responsiveness to such conditions may be fundamentally different
from what is nowadays called conformity.

4. We finally come to constraint and conformity proper. These high-
light another aspect of social determination. They refer to conditions
that create a conflict between tendencies in the person and the forces
extending from the social field. They differ from the instances men-
tioned previously in that they present an issue and involve a choice
among alternatives. As soon as this is the case, the happenings are a
function of the conflicting alternatives. The problems in this region con-
cern the operations of conformity and independence, not of conformity
alone.3

Current thinking has assigned a particular interpretation to the con-
straint-conformity operations. It relies on a general proposition about re-
wards and punishments, derived from observations of lower organisms,
and silently assumes the egocentric axiom which was discussed earlier.
It also presupposes that constraint and conformity are psychologically
homogeneous. Actually there is a crying need to discriminate among dif-
ferent kinds of constraint and conformity.

Prior to formal investigation we can enumerate a few.

1. There is a range of constraints that persons accept because they
seem just, even if not pleasurable. If so, it follows that the study of
social influences cannot be pursued at some crucial points in the absence
of a psychology of ethics. Current thinking and investigation has ignored
this fundamental basis of social discipline. The belief that the vectors
present in the experiences of right and wrong are merely the reflections
of social influences, a view that is not wholly intelligible, has silenced
questions that should be raised.

2. Action in line with social demands, even if their lightness is in
doubt, is, of course, a fact of considerable importance. It is customary
to refer here to the role of expediency, but it may be more illuminating

2 The forms of social determination described above need not, of course, occur
in separation. They may be, and probably often are, relevant to the same set of
events.

382 SOLOMON E. ASCH

to consider the varieties of motivation that are implicated, (a.} Fear of
consequences, the favorite formula for the explanation of social sanctions,
is a potent force. But It may be necessary to strip It of Its simplicity. It Is
pertinent to ask why the anticipation of punishment is effective in some
circumstances, not in others, with some persons and not others, (b. >
Loyalty to the group. A worker may be convinced that a call to a strike
is unwise, but will lay down his tools because he believes that the welfare
of his union will be best served by his acquiescence. This quite human
and powerful attitude seems not to have found credence in our psy-
chology, (c.) Another potent source of conformity is indifference, the
failure to see an issue, and the pressure of other concerns which are
presumed to be of higher importance. (d.)One may even conform in
order to exploit others for one's own ends.

The preceding examples have not included those effects that Involve
a change of evaluation of the given data. They have left out of account
the power of social conditions to alter established judgments and con-
victions. Again it is unlikely that a single abstract formula will do justice
to operations that range from simple cognitive inference to the most
complicated changes of emotions and attitudes. Inquiry will need to find
a place for ( 1 ) the properties of narrowed mental fields, ( 2 ) the sources
of respect for authority in matters remote from Immediate experience,
and ( 3 ) the dependence of distortions in feeling and thinking on the need
to preserve cherished personal and group bonds. Since our understanding
of these matters is limited, qualitative observation has a place of Im-
portance.

In this region, too, a certain breadth of view is not out of place.
Thinking and investigation have concentrated almost obsessively on con-
formity in its most sterile forms. Observation of human affairs, as well
as psychological considerations, can correct this one-sidedness. The indi-
vidual participates in social life by means of his capacities to think and
feel, by including within his view the situation of the group. Individuals
stand in a relation to their group milieu wholly different from that of a
cell to an organ, or an organ to the body. They will never be free of group
constraints, but they are potentially capable of questioning the most
established beliefs. To be sure, if one takes a sufficiently narrow — or
overextended — view, the majority of mankind appears throughout history
as an inert mass swinging heavily with the social tides. The notions of
imitation and conformity then seem to fit most aptly. But it is the con-
tribution of psychological thinking to take up a position that is neither
too near nor too far from its subject matter. We may agree that the
greater part of mankind takes most of its ideas and beliefs at second
hand, that the reasons for most of the things men do is that others have
done them* But the psychologist will not miss seeing that in some corner

A Perspective on Social Psychology 383

of our lives we are at times capable of taking a fresh view, and that the
aspiration to become oneself is ako part of the human attitude.

REFERENCES

1. Allport, F. H. Social psychology. Boston: Houghton Mifflin, 1924.

2. Asch, S. E. Social psychology. Englewood Cliffs, X.J. : Prentice-Hall,
1952.

3. Bollard, J., Doob, L. W., Miller, N. E.3 Mowrer, O. H., & Sears,
R. R. Frustration and aggression. New Haven, Conn.: Yale Univer. Press,
1939.

4. Heider, F. Social perception and phenomenal causality. Psychol Rev.,
1944, 51, 358-374.

5. Hull, C. L. Principles of behavior. New York: Appleton-Century-
Crofts, 1943.

6. Koch, S., & Hull, Clark L. In Estes, W. K., et al. Modern learning
theory. New York: Appleton-Century-Crofts, 1954.

7. Kohler, W. Gestalt psychology. New York: Liveright, 1929.

8. McDougaJl, W. The group mind. New York: Putnam's Sons, 1920.

9. Sears, R. R. A theoretical framework for personality and social be-
havior. Amer. Psychologist, 1951, 6, 476-483.

10. Skinner, B. F. Science and human behavior. New York: Macmillan,
1953.

11. Skinner, B. F. Walden two. New York: Macmillan, 1948.

INDIVIDUAL SYSTEMS OF ORIENTATION

THEODORE M. NEWCOMB
University of Michigan

Introduction {!> 384

Concepts essential to the formulation {2J 388

Orientation 388

Systems of orientation 392

System strain 393

Communicative behavior 394

Major Interrelations among Constructs {2+} 395

Certain states of orientation systems are strain inducing 397

Instigation to communication is a learned response to strain .... 400
Under certain conditions strain reduction follows communication . . .401
Under certain conditions strain -reducing system change occurs in the ab-
sence of overt communicative behavior 402

Alternative intrasystem changes may have equivalent effects upon strain . 403

Alternative Formulations {?)- 406

Evidence Relevant to the Present Formulation {8, 9} 408

Samples of relevant evidence from published studies 41 0

The interdependence of frequency of interaction with positive attraction

toward other persons 41 0

The interdependence of frequency of interaction and objective similarity of

attitudes 411

The interdependence of frequency of interaction and perceived similarity of

attitudes 412

Personal attraction and objective similarity of attitude 413

Personal attraction and perceived similarity of attitude 413

Objective similarity and perceived similarity of attitude 414

History and Prospects of the System in Mediating Research {8, 11, 12} . . 416

References 420

INTRODUCTION

This paper* attempts to explore, both theoretically and empirically,
some of the consequences of a single postulate. In its most general form,

* Prepared while the writer was a Fellow at the Center for Advanced Study
in the Behavioral Sciences, Stanford, Calif., in 1956-57.

384

Individual Systems of Orientation 385

the postulate is to the effect that there are lawful interdependencies
among certain classes of beliefs and attitudes held by the same individual.
Proceeding from distinctions between two classes of objects of attitudes
( persons as communicators, and objects of communication), and be-
tween "own53 attitudes and those attributed to other persons, it hypothe-
sizes that certain kinds of combinations of such attitudes and beliefs are
psychologically unstable, tending to induce psychological events that re-
sult in more stable combinations. A more precise wording of the proposi-
tion must await further clarification of terms.

I know of no better way to indicate the kinds of problems that I have
found illuminated by the approach here outlined than to trace its natural
history, autobiographically.

In one of my earliest investigations [35] I found, not very surpris-
ingly, a very considerable degree of within-family homogeneity in atti-
tudes toward political and religious issues. What I had not anticipated —
after all, my training had been in psychology, not in sociology — was the
finding that the variations in degree of within-family similarity could
only be accounted for in terms of differential impacts upon different
families of common influences, which seemed to be of institutional na-
ture. It seemed likely that individual differences within families whose
members were subjected to common institutional impacts represented
some sort of compromise adaptation to family norms, on the one hand,
and to personal attitudes toward family members, on the other.

In a later and more ambitious study [31] I found some confirmation
for these suspicions. In tracing the development of attitudes toward pub-
lic issues of the entire student population of a small college, the influence
of group norms was very clearly revealed, but as so often happens, it was
by careful examination of variations on the common theme, and of out-
right exceptions to it, that the most significant findings emerged. As a
general principle, attitude change in the direction of the prevailing norms
was most pronounced on the part of those who participated most actively
in college affairs. But change and persistence in attitudes toward public
issues could be accounted for in all cases only in the light of attitudes
toward persons and groups. Somehow the two kinds of attitudes were
associated. Stimulated in particular by discussions with Professor
M. Sherif, I published an account [33] of these findings, subsequent
to the original monograph, in terms of positive and negative reference

groups.

No findings had been more illuminating, in this investigation, than
those derived from subjects5 estimates of the attitudes of various persons
and groups toward the same issues concerning which their own attitudes
had been frequently expressed. The degree and direction of distortion
that appeared in these estimates appeared to be a particularly sensitive

386 THEODORE M. NEWCOMB

indicator of the psychological processes by which approach-avoidance
tendencies toward persons or groups influenced acceptance or rejection
of the positions attributed to those persons or groups. Conversely, changes
in attitudes toward persons and groups often accompanied changes in
attitudes toward public issues, under conditions suggesting that the
former was the dependent variable.

From such considerations it was not much of a theoretical leap to
speculate along the following lines. Human beings are constantly de-
pendent upon each other, not only in direct ways and for consummatory
purposes (like helping and loving) but also indirectly, as sources of in-
formation about other objects in the world. One's own eyes and ears are
never enough; one's own experience is almost always too limited, and
one's own observations often in error. Socialized1 human individuals are
as rewarding to each other in the latter sense — i.e., as communicators, as
suppliers of new information, and as confirmers or correctors of old — as
in the former. Favorable attitudes toward persons, as rewarders, are
generated in both ways.

In so far as one person has been rewarded by another as com-
municator— i.e., rewarded in the sense of having found him a trust-
worthy informant — he is likely to experience conflict on discovering that
his own attitudes toward some object are divergent from those of the
trusted informant. In somewhat more general terms, it may be postulated
that perceived discrepancy between own attitude and that of a trusted
person or group is disturbing; it serves to upset a previously established
equilibrium. The discoverer of such a discrepancy is likely to conclude
that either his own attitude or that of the previously trustworthy in-
formant is "wrong.35 Such a disturbance might be expected, as in the
case of other equilibrium disturbances, to engender some form of equi-
librium-restoring behavior — for example, changing one's own attitude
toward the object, obtaining further information from other sources, or
modifying one's trust in the informant.

My thinking along these lines was considerably aided by Professor
F. Heider9s published work [13, 14] on "balance," and later by that of
my colleagues Dorwin Cartwright and Frank Harary [6], along similar
lines. Meanwhile the experimental and theoretical contributions of
Professor L. Festinger and his students [see especially 10, 11] helped
me to put into perspective the interrelated notions of perceived dis-
crepancy in attitude and communicative behavior.

As a result of these and other influences I have ventured into some-

1 Here, and throughout this paper, I shall use the term "socialized" to refer to
humans old enough and otherwise able to communicate "normally" and to have
"internalized" the norms of groups of which they are members sufficiently not to
be considered gross deviants.

Individual Systems of Orientation 387

what wider explorations concerning social interaction, viewed from the
point of view of both groups and individuals. I have come to see social
interaction in communicative terms, in the sense that it involves almost
exclusively the exchange of information rather than of energy. I view
individual participation in social interaction as a virtually lifelong
process which includes the following subprocesses :

1. Cognitive and cathectlc predispositions (attitudes, or orientations,
as I shall later call them) are acquired, interdependently, toward persons
and toward objects of communication with those persons.

2. Simultaneously, beliefs are acquired concerning the attitudes of
fellow-communicators toward objects of communication with them.

3. With regard to specific persons (or groups) and specific objects
(or classes of objects) of communication, attitudes and beliefs about
others' attitudes come to function interdependently as a system having
equilibrium properties.

4. Communicative exchange is initiated by individuals under condi-
tions of system disequilibrium,

5. System modifications tending toward restored equilibrium follow
such communicative exchanges.

6. The new state of equilibrium tends to persist until it is disturbed
by the receipt of new information (by direct, sensory experience with the
object, by communication with others, or by processes of memory,
reasoning, fantasy, etc.), following which there is renewed instigation to
communicative exchange.

In an immediate sense, this paper attempts little more than a
systematic formulation of the processes by which human beings develop
attitudes toward other persons and toward objects of joint relevance to
themselves and to those persons. I believe that the formulation takes into
account a wider range of phenomena, and brings them more par-
simoniously within a single framework, than have my own previous at-
tempts in this direction. In a more inclusive sense, I have some hope that
the approach here presented has improved my own understanding of the
peculiarly human aspects of social interaction in all its forms. Perhaps a
similar formulation, in collective rather than in individual terms but
resting upon the same general notions, can even be applied to the
understanding of group phenomena at their own level — but that is a
different story, more appropriately told elsewhere. For the present, I
need only note that "systems of orientation," to a fuller description of
which I now turn, may be regarded as intraindividual representors of the
objective interactional systems in which individuals are psychologically
involved at any given moment.

At the outset I stated my intention of exploring the consequences of
a single postulate, a formal statement of which depends upon terms yet

388 THEODORE M. NEWGOMB

to be presented. That postulate is to the effect that forces toward states

of equilibrium within individual systems of orientation determine both
existing attitudes toward two classes of objects and the behaviors by
which further information concerning those objects is obtained and
evaluated — and by which, therefore, those attitudes are maintained or
changed. The consequence of taking this proposition seriously is the
promising possibility of developing an inclusive theory which, at one
level, accounts for the development of multiple attitudes on the part of
single individuak and, at another level, accounts for the communicative
behavior of which social interaction among humans so largely consists.

CONCEPTS ESSENTIAL TO THE FORMULATION

All the concepts described below refer, in a sense, to independent,
intervening, or dependent variables, since each of them refers to some-
thing conceived as varying in degree, and since each of them, hypo-
thetically, either contributes to the variance of one or more of the others
or results from such variance. For two reasons, however, I have preferred
to present this formulation in terms of systematic rather than empirical
variables. First, the heart of the formulation lies in hypothetical rela-
tionships of the several variables to the construct "strain,55 which is in no
sense an empirical variable. Since, as I assume, an empirical variable is
meaningless apart from a proposition in which it is paired with at least
one other empirical variable, either as dependent or as independent, it
would be inappropriate to present this formulation as a set of hypo-
thetical relationships among empirical variables. Second, I have chosen
to emphasize "system properties" rather than the single variables which
contribute to them, and consequently none of the variables has an en-
during status either as independent or as dependent. According to some
of the specific propositions to be presented, a change in one system
variable is likely (under certain conditions) to be followed by a specified
change in another system variable, but according to others a change
in the second is a precondition for a change in the first. The variables
corresponding to the following concepts are therefore presented as formal
ones, which may or may not be subject to operational definition, but
whose hypothetical effects are empirically testable.

The fact that I have chosen to emphasize the systematic nature of
these construct variables will not, I trust, lead the reader to conclude that
the present formulation has not led to testable hypotheses. The facts
are quite the reverse, as I shall attempt to show, after a fuller description
of the central concepts, and of the relationships among the systemic
variables to which they correspond.

Orientation. It seems to me necessary to assume that human in-
dividuals, in interacting with one another, develop cognitive and

Individual Systems of Orientation 389

cathectic habits of relating themselves both to each other and to the
world of objects with which they deal in common. So dependent are
human beings upon one another, and so recurrent are the requirements
for adaptation to the objects which they face in common that, given
human capacity for learning, the development of habitual and antic-
ipatory adaptations to one's fellows and to common objects is inevitable.
Since it is the cognitive and the cathectic aspects of these habitual
adaptations that are of interest in the present formulation, the term
"attitude/3 which commonly has precisely such meanings, would serve
to describe them, except for one consideration. It is crucial for the
present formulation to distinguish between twro kinds of objects of
attitudes: persons as fellow communicators, and objects of communica-
tion (including persons). I have therefore preferred to use "orientation55
inclusively, as referring to both kinds of objects of attitudes, reserving
for "attitude53 the latter meaning only and labeling the former as "at-
traction."

Conceptually, an orientation may be defined (in its most general
sense) as that existing organization of the psychological processes of an
organism which affects its subsequent behavior with regard to a dis-
criminable object or class of objects.2 By long usage, however — doubtless

2 This resembles fairly closely the "standard" definitions of "attitude." Cf.
Kretch and Crutchfield: "an enduring organization of motivational, emotional,
perceptual, and cognitive processes with respect to some aspect of the individual's
world"; or Newcomb: "predisposition to perform, perceive, think and feel in re-
lation to" an object; or G. W. Allport: "a mental and neural state of readiness,
organized through experience, exerting a directive or dynamic influence upon the
individual's response to all objects and situations with which it is related." The
inclusion in the definition of the notion of persistence over time is in one sense
very awkward, since if the behavior from which an attitude is inferred changes
over time, it becomes necessary to assume that the attitude defined as "enduring"
has not literally endured, but has changed. Nevertheless, there would be no need
of the concept of attitude if the time factor were not taken into account, because
without it the concept would refer only to momentary determinants of behavior.

The source of the dilemma lies in the temptation to assume that object-oriented
behavior is determined exclusively by attitudes, overlooking the fact that there
are also immediate situational determinants. Any given instance of object-oriented
behavior is a resultant of attitudinal ( presituational, residual from previous ex-
perience) and of immediate (situational) determinants. Instead of "enduring,"
therefore, I have used the term "existing," intending to suggest (a} that attitudes
consist of the presituational determinants of any given instance of behavior with
regard to a specified object, or class of objects; and (b) that if behavior in a
given instance is not as would have been predicted from knowledge of presitua-
tional attitude, it is not necessary to assume that the "persistent" attitude has
changed (as known only post hoc) but only that new, situation-induced influences
have been introduced. These new influences, from the present point of view, may
subsequently modify the attitude, but these determinants of the specific instance of
behavior are not to be included among the attitudinal determinants of that be-
havior.

390 THEODORE M. NEWCOMB

because of the ways in which the concept has in fact been operationalized
— its meaning has been pretty much limited to "strength" and "di-
rectionality" or sign (i.e., approach-avoidance) as aspects of the "exist-
ing organization . . . which affects . . . subsequent behavior.55 The
narrower conceptual definition would thus become "that existing organ-
ization of the psychological processes of any organism which affects the
direction and strength of its subsequent behavior with regard to a
discriminable object or class of objects.53

Orientations are here categorized in twro ways: according to the role
of the object of orientation in the communicative process (attitudes and
attractions), and according to psychological aspect (cognitions and
cathexes). Attitudes are conceptually defined as orientations toward any
object viewed as object of communication — that is, about which in-
formation may be transmitted and/or received.3 Attractions (which, like
attitudes, may have either plus or minus sign), analogously, are con-
ceptually defined as orientations toward cocommunicators — specifically,
toward the source of a message, on the part of the receiver (actual or
potential4), or toward the recipient, on the part of the transmitter.

This distinction, however simple it may appear conceptually, has as
an empirical counterpart an important class of borderline orientations;
namely, those in which the object of communication is either the source
or the recipient of the message. There are many instances in which it is
possible to distinguish empirically between persons as objects of com-
munication and as sources or recipients of communication about some-
thing other than themselves. Thus, the transmitter of the message
"Lincoln was a wonderful man" is not the object but only the author of
the communication. But if he transmits the message, "I am hungry," he
is both the author and object of the communication. In cases of the latter
kind it is empirically difficult to distinguish between the orientations of
the recipient of the message to the transmitter qua transmitter and
qua object of communication, i.e., between the receiver's attitude and
his attraction toward the transmitter. This difficulty is not merely one
of operations, but seems to rest upon the solid empirical facts of psy-
chological generalization; that is, persons frequently do not in fact keep

8 As implied by the phrase "and/or," I shall not limit the term "communica-
tion" to those instances of message transmission in which both encoding and de-
coding occur. Though it is possible to treat messages which are not, in the literal
sense, encoded (because they are unwittingly transmitted) as signs rather than as
messages, I shall consider the occurrence of such events as communicative phenom-
ena, provided they are receivable and decodable messages. This seems to be con-
sistent with Miller's definition: " 'Information' is used to refer to the occurrence
of one out of a set of alternative discriminative stimuli" [28; my italics].

4 Henceforth, in the interests of brevity, I shall not repeat the phrase "actual or
potential," which will be taken for granted.

Individual Systems of Orientation 391

distinct their orientations toward communicators and toward the same
persons as objects of communication. Hence, although the conceptual
distinction is clear enough, it is not always possible to make the empirical
distinction.

Orientations have been categorized, secondly, as having both ca-
thectic and cognitive aspects. The former refer to approach-avoidance
tendencies; cathectic orientations have the conceptual properties of
sign and strength. It is convenient (and traditional) to regard sign and
strength as varying along a single continuum, from maximally positive
to maximally negative approach tendencies.

Cognitive orientations (or, more exactly, the cognitive aspects of
orientations) have to do with the ordering, or structuring, of attributes
as cognized "in" the object of the orientation. For present purposes it is
not assumed that orientations vary in respect to degree of "ordering
of attributes," but only in respect to the relative salience of specified
attributes. Such attributes, which may themselves be regarded as sub-
objects of orientation — i.e., aspects of the "whole" object — may also have
cathectic value, but if so it is the phenomenal ordering, or relative
salience, of these attributes, and not any central tendency of their
cathectic values, which is the important property of cognitive orienta-
tions, for my purposes. It is discrepancies among cognitive orientations —
in particular on the part of different individuals toward the same ob-
ject— and not an individual's cognitive orientation toward an object,
that constitute a system variable. Cognitive orientations toward different
objects, or toward the same object by different persons, are comparable
only in so far as the same attributes are ordered. They are not com-
parable if only the attributes peculiar to each object of orientation, or if
only those recognized by a given person, are taken into account.

Operationally, cathectic orientations (whether attitudes or attrac-
tions) are ordinarily defined in terms of verbal responses from which sign
and strength are inferred ; they may also, of course, be inferred from non-
verbal behavior. Any of the "standard" procedures of sociometric or
attitude measurement may be employed. Cognitive orientations may be
operationalized from verbal responses like checking the presence or
absence of attributes of objects, or rank-ordering attributes on a con-
tinuum of salience. I have found Gough's adjective check lists [12]
useful for the study of cognitive aspects of persons.

Judged (or, less accurately, perceived) orientations of others refer
to any of the above categories of orientations, as attributed to another
person, or, for certain purposes, to a group. As will subsequently appear,
it is the relationships between subjects' own orientations and those which
they attribute to others, toward the same objects, that are crucial for the
present formulation, rather than either kind alone. For the sake of com-

392 THEODORE M. NEWCOMB

parability, therefore, it is important that judgments of others' orienta-
tions be obtained via the same instruments by which ccown33 orientations
are indicated.

Systems of orientation. The fundamental postulate upon which the
present formulation rests is that all of the foregoing kinds of orienta-
tions, and of judged orientations of others, on the part of the same in-
dividual (assuming that all of them may be said to exist) are dynamically
interdependent. (The argument upon which this postulate rests is
presented in the following section.) The totality of these orientations is
therefore regarded as having system properties, in the sense that a
change in any one of them, under certain hypothetical conditions, in-
duces change in one or more of the others. Orientations and judged
orientations of others are regarded as elements in such systems, rather
than as variables of primary significance in their own right, although

, •< ~ g FIG. 1. Schematic representation of orientations in-

" eluded in the phenomenal system of person A, as he
co-orients toward person B and toward object of
communication X. (Arrows point toward the person
or object of orientation; broken lines refer to ori-
entation of person B as perceived by person A, and solid lines to person A3s own
orientation.)

important propositions may be derived in which orientations appear as
independent or as dependent variables.

Figure 1 presents a schematic illustration of the system of orienta-
tions of individual A with respect to individual B and object X; the
arrows refer to both cognitive and cathectic aspects of orientations. A is
said to be co-orienting toward B and X when all the orientations
represented by the arrows in the figure are phenomenally present in A.
I have previously [30] indicated my reasons for believing that co-
orientation is the rule rather than the exception. On the part of socialized
humans, that is (with the possible exception of utterly private objects),
orientations toward objects are never unaccompanied by orientations to-
ward other persons who are assumed also to have orientations toward
them, and orientations toward persons are never unaccompanied by
orientations toward objects toward which they are assumed also to have
orientations.

There are other orientation variables which function primarily as
parameters of system strain. Importance refers to the valence aspects of
attitudes, or to degree of cathectic orientation toward an object, regard-
less of sign. It is most readily operationaHzed in terms of an intensity
measure, or in terms of degree of positive or negative attitude. Object
relevance refers to the degree of joint dependence of two or more com-

Individual Systems of Orientation 393

municators upon a specific object of communication, as judged by one of
them. Or, alternatively, it is the strength of forces acting upon an in-
dividual to co-orient toward a specified object and toward a specified
other person. It is operationalized from verbal responses concerning the
importance of the object to the self-other relationship, or concerning the
closeness or the relative frequency with which the respondent associates
the object with the other person. Thus a child would be a highly relevant
object to each of its devoted parents, provided that each of them assumes
that the other shares his or her concern for the child. The business
affairs of a man who considers that his business is "not the business" of
his wife, and that she is indifferent to them, would not, from his point
of view, be an object of joint relevance to both.

Certain distinguishable forms of attraction (liking, respect, trust)
are described below, in context; fuller statements about them do not
seem necessary meanwhile.

System strain. The nature and the conditions of interdependence
among orientations and judged orientations of others depend upon the
hypothetical intervention of an additional construct — that of "strain."
In an earlier formulation [30] I referred to this construct as "strain to-
ward symmetry" — a phrase whose suitability derived only from a certain
graphic presentation of the elements of systems of orientation. The
phrase is not an altogether happy one, but in any event it refers to a
hypothetical state of psychological tension occurring under certain condi-
tions of judged discrepancy between own and another's attitude toward
the same object. As outlined in greater detail below, systems tend to
move from states of greater to lesser strain, or toward "balance" [cf. 13,
6]. The mechanism by which this is accomplished is that of change in
one or more of the system elements (orientations and judged orientations
of others). This commonly but not necessarily occurs following com-
municative behavior, to which system strain is hypothetically an in-
stigator.

Strain is regarded as corresponding to a state of tension [in its
Lewinian sense; cf. 24, chap. 1] induced by the judged state of the
cocommunicator's orientations in relation to one's own. The source of
such tension may be ( 1 ) perceived discrepancy of self-other orientations
and/or (2) uncertainty as to the other's orientations. The distinction is
necessary because either may occur without the other, and their con-
sequences may be very different (In everyday terms, one may need to
know another's orientations without in the least caring how they cor-
respond to one's own — e.g., "How will you take your tea?" ; or one may
know with great certainty that another's orientations are different from
one's own and be greatly concerned about the discrepancy.) The
hypothetical conditions under which both discrepancy strain and un-

394 THEODORE M. NEWCOMB

certainty strain are aroused are discussed below. Meanwhile, it need
only be said that it is not assumed that either perceived discrepancy or
uncertainty, per se, necessarily involves strain.5

Amount of strain, as a system variable, is postulated to vary with
( 1 ) degree of perceived discrepancy, ( 2 ) sign and degree of attraction,
(3) importance of the object of communication, (4) certainty ("com-
mittedness35) of own orientation, and (5) object relevance. The nature
of these functional relationships is discussed below.

System-strain variables are categorized as accompanying either posi-
tive or negative attraction. This distinction is an important one, the-
oretically, since the manner in which attraction variables are presumed
to interact with other variables to produce a given amount of strain
varies with sign of attraction. It is convenient, therefore, to speak of
positive and negative strain, as varying with the sign of attraction to-
ward the cocommunicator. The further distinction between cognitive and
cathectic strain corresponds to phenomenal discrepancies between the
cognitive or cathectic aspects of orientations. Thus, for example, posi-
tive cathectic strain would be said to characterize any state of a system
of orientations in which attraction to the cocommunicator Is positive and
in which there is perceived discrepancy between own and other's cathec-
tic orientations to a relevant object of communication.

Communicative behavior. For present purposes, communicative be-
havior consists of the transmission and reception of information by
human organisms. Information has been formally defined by Miller as
"the occurrence of one out of a set of alternative discriminative stimuli"
[28]. In the case of humans, at least, the latter are necessarily symbols
(though not necessarily verbal ones) . Hence, behaviorally speaking, com-
munication, as I shall use the term, consists of sending and receiving
symbols.

Communicated messages may be categorized (1) as consisting of
either transmitted or received information, and ( 2 ) as to the content of
the information. Since the central theoretical problem of the present sys-
tematic formulation has to do with the interdependence of communica-
tive behavior and the orientations of the communicators, the primary
content variables are the inferred orientations of the transmitter toward
the referent of the symbols used in his communication.6 These are sub-
categorized, again, as either cathectic or cognitive. Inferences about the

5 Unless "uncertainty" is specifically mentioned, "strain" will henceforth refer
to "perceived discrepancy" and not to "uncertainty."

6 For theoretical purposes, it is, of course, fruitless to categorize content in terms
of objects of communication (symbol referents), since such categorizations would
be phenotypic and not subject to theoretical generalization. Categorization in terms
of inferred orientations of the transmitter has the common-sense justification that
behavioral consequences for the receiver vary more directly with his inferences

Individual Systems of Orientation 395

transmitter's orientations may of course be made either by the "objec-
tive" observer or by the recipient of the message; such inferences are
assumed to influence the recipient's subsequent behavior in relation to
the transmitter or to the object of orientation, or both.

There are two systematic variables closely related to the notion of
communicative behavior. One of these is "amount of information re-
ceived55 by a given individual from another given individual about a
given object of communication. For obvious reasons it would be virtually
impossible to apply the conventional measure (the logarithm of the
number of alternatives) in the more or less "natural53 situations In which
investigations suggested by the present formulation could be carried out.
My own practice, therefore, has been to use very crude Indexes, like
amount of time spent in discussing a given topic with a given individual.
The other systematic variable, "instigation to communicate,53 refers to a
hypothetical degree of motivation to transmit a message to, or to initiate
a communicative exchange with, a given individual about a given object.
Such a construct is necessary, as noted below, because there are many
possible counterforces which result in inhibition of communications that
would otherwise, presumably, be transmitted.

MAJOR INTERRELATIONS AMONG CONSTRUCTS

The general outlines of the relationships among orientations, system
strain, and communicative behavior have already emerged. These rela-
tionships are conceptualized as properties of an intrapersonal system of
orientations, and of perceived orientations of another person, on the part
of a co-orienting person (i.e., one who is attending both to a cocommuni-
cator and to an object of communication) .

More explicit statements are now in order; at the broadest level of
generality they will be formulated as inclusive postulates from which
more specific, testable hypotheses may be derived.

1. Certain combinations of a person5s orientations toward a specified
object of communication and toward a cocommunicator, together with
the Iatter3s perceived orientations toward the same object of communica-
tion (all of which orientations are viewed as constituting a system), are

about the transmitter's orientations than with the "pure" content of the com-
munication. It has the theoretical advantage, moreover, of making minimal as-
sumptions (e.g., the message "I like apples" does not necessarily have the conse-
quence for the receiver of the message that he believes that the transmitter likes
apples, nor if he does, that thereafter he considers apples good), and thus leaving
open the question of other parameters involved in the behavioral consequences for
the receiver. Such a categorization has the disadvantage, of course, of necessitating
very difficult operationalizations.

396 THEODORE M. NEWCOMB

strain inducing. For example, if a student discovers that some of his be-
liefs about genes are at odds with those of his respected professor of
genetics, or if he believes that a close friend differs with him about the
virtues of a favorite political candidate, both systems of orientation
would be characterized by discrepancy strain.

2. Instigation to communicative behavior is a learned response (not
necessarily the only one), on the part of socialized humans, to such
strain. The likelihood that such instigation will actually eventuate in
communicative behavior (e.g., asking questions, trying to persuade the
other to one's own point of view) varies with the strength of instigation,
situatlonal opportunity for communication, and the strength of opposing
influences. Thus the student of genetics is most likely to seek further in-
formation from his professor if he is much disturbed by the discrepancy,
if the professor is readily accessible, and if the student is not afraid, em-
barrassed, or otherwise reluctant to launch the communicative exchange
to which he is instigated.

3. Following communicative behavior (transmitting and/or receiving
information) on the part of one person vis-a-vis another, changes may
occur within his system of orientations such that strain is reduced. The
probability with which this occurs varies both with internal system
variables (e.g., attraction toward the other, or degree of perceived dis-
crepancy) and with external parameters (e.g., "competing" attraction
toward other persons or groups) . It is much more apt to occur, of course,
following the receipt of information (especially concerning another's
orientations) than following its transmission without feedback from the
other person. If the same student has involved his friend in a discussion
about the political candidate in question, his discrepancy strain might
be reduced in any of the following ways: by being influenced to change
his own attitudes toward the candidate; by becoming convinced that his
friend's opinions had changed in the direction of his own; by concluding
that his previous assumptions about his friend's opinions had been in
error; by concluding that the matter was really of very little importance;
or by concluding that his friend was so incompetent with respect to
politics that his opinions did not matter. The first of these would be least
likely to occur if the student's committedness to his own point of view
was very strong; the last would be least likely to occur if his attraction
toward his friend was very strong, and if there were many objects of
importance to both of them which did not involve discrepancy strain.

4. The interdependence relationships within systems of orientation
are such that, under conditions of system strain, changes in one or more
of the component orientations or judged orientations may result in strain
reduction apart from any overt communicative behavior. Suppose that
the student prefers not to engage in what he fears may turn out to be an

Individual Systems of Orientation 397

unpleasant argument with Ms friend about politics. System strain is likely
to be reduced intrapersonally, or autisticaily, in such ways as to have the
same consequences as those noted above as following communication.
That is, he may persuade himself that his own attitudes are too extreme;
or that he must be in error about his friend; or that the matter is trivial,
and should be shrugged off; or that Ms friend, though a fine fellow,
must be considered a bit of an innocent in the area of politics. It is postu-
lated, however, that these are substitutes for overt communication in the
presence of counterforces to communicate, and not substitutes for the
initial instigation to communicate.

5. The interdependence relationships within systems of orientation
are such that alternative intrasystem changes may have equivalent effects
upon strain. Eventually, that is, systems of orientation tend to revert
toward equilibrium (i.e., minimal strain), whether the initial instigation

States of systems
of orientation

FIG. 2. Schematic illustration of relationships among essential concepts.
Under certain conditions at the source of each arrow, certain changes
hypothetically occur in the phenomena toward which the arrows point.

to communicate is expressed or inhibited, and whether one or another or
some combination of the various alternatives is employed.

The first four of these statements of relationships among existing
states of systems of orientation, communicative behavior, and changes in
states of systems are schematically illustrated in Fig. 2. Each of the
arrows indicates that "under certain conditions so-and-so occurs or is
likely to occur."

There follow somewhat expanded explications of these five postulates.
At the risk of repetitiveness, I have preferred to present all of them
briefly before making the fuller statements, in order that each of the
latter may be understood in the light of preliminary familiarity with all
of them.

Certain states of orientation systems are strain inducing. The most
unambiguous instance is the combination of positive attraction with per-
ceived discrepancy of attitude toward a relevant object. Such a system
state involves conflict or threat, and is, therefore, tension inducing, in
one or more of the following ways:

1. In so far as positive attraction involves respect for the other's
knowledgeability or expertness regarding the object of communication,
an orientation which diverges from the other's may be in error. Hence

398 THEODORE M. NEWCOMB

the greater the respect, and the greater the Importance of not being in
error, the greater the strain (degree of discrepancy being held constant).

2. In so far as positive attraction involves trust in the other's willing-
ness to be helpful with respect to the object of communication, a di-
vergent orientation may threaten motive satisfaction with regard to that
object, since cognitive discrepancy might interfere with the communica-
tion through which help is to be given and received, and since cathectic
discrepancy might lead to unwanted kinds of "help." Hence, given posi-
tive respect (i.e., perception of the other's ability to be helpful, without
which trust becomes irrelevant) , the greater the trust and the greater the
object-relevance, the greater the strain of discrepancy (i.e., the greater
the assurance that the other is both able and willing to help, and the
greater the need of that help, the more threatening perceived discrepancy
becomes).

3. In so far as positive attraction involves generalized liking for the
other person (together with motivation to associate with him), a diver-
gent orientation might threaten the personal relationship, either via overt
conflict or (if the other's tolerance for discrepancy is thought to be low)
via the threat of rejection by him. Hence the greater the liking, the
stronger the person's own committedness, and the stronger the other's
perceived committedness, the greater the strain, other things equal.

4. And finally, under all conditions of positive attraction, regardless
of its particular components, sheer uncertainty as to whether or not dis-
crepancy of orientation exists may be threatening; behind uncertainty
there may be the possibility of any of the kinds of threats mentioned
above. Beyond a certain point, presumably, known discrepancy may be
more tolerable than uncertainty; and below the point where other strain-
inducing conditions exist, uncertainty would not be threatening. Within
these limits, the greater the uncertainty as to divergence of orientations
the greater the strain, ceteris paribus.

In general, the conditions of strain induction are the same for nega-
tive as for positive attraction. Such differences as there are between posi-
tive and negative strain are analogous to the differences between any
instance of simple approach behavior and the same behavior as avoid-
ance of a less attractive alternative.

An exact analysis of the conditions of negative strain presupposes an
analysis of the conditions under which co-orientation toward a negatively
attractive person (together with an object of communication) occurs.
In particular, it is necessary to distinguish between those situations in
which co-orientation occurs because of negative attraction toward the
cocommunicator and those in which it occurs in spite of negative attrac-
tion. In the former case, it is perceived threat potential of the cocom-
municator which induces co-orientation toward him and toward objects

Individual Systems of Orientation 399

related to his feared behaviors. Under these conditions, as with positive
attraction, the greater the negative attraction, the greater the strain (i.e.,
the more dangerous the other person, the greater the Importance of being
able to see things as he sees them). Co-orientation may occur in spite of
negative attraction either because of perceived object-relevance (e.g.,
toward a business client who Is held in contempt and toward the trans-
action that must be completed with him), or because of conditions that,
for our purposes, are extra-systematic (as In conversation about the
weather with a bore who cannot be avoided). Under these conditions,
as distinct from positive attraction, there is no reason to hypothesize that
strain increases with degree of negative attraction, but as with positive
attraction, strain Is hypothesized to increase with object-relevance.

Certain differential predictions which are made for the several
"varieties" (which really refer to "sources") of attraction correspond
fairly closely to those just made for the two signs of attraction. Respect
is conceptualized as "adlence toward the cocommunicator stemming
from his perceived power over the object of communication," including
knowledgeability, skill, expertness, and ability to make decisions about
It; respect is object-specific. No theoretical purpose appears to be
served by conceptualizing respect as negative — i.e., less than no skill,
expertness, etc. Its contributions to strain are, therefore, hypothetically
the same as for other varieties of positive attraction. Trust is specific
to the co-orienting person (i.e., to the self), just as respect is specific
to the co-oriented object; it is conceptualized as "an Individual's
adience stemming from the cocommunicator's perceived favorableness
toward that individual," including sincerity and helpfulness on the
positive side and deceitfulness and hostility, on the negative. As sug-
gested by the just preceding discussion of negative attraction induced by
threat potential, degree of strain varies directly with strength of trust,
regardless of sign.

Liking is conceptualized as "general, undifferentiated adience (or
abience, in the negative case) toward the cocommunicator"; its sources
are not specified. It may be presumed to be a generalized resultant of the
system properties of respect and trust, and of properties attributed to
the cocommunicator which, for present purposes, may be regarded as
extrasystematic. As in the case of trust, strain is presumed to vary
directly with degree of liking, regardless of sign.

Strain, so far, has referred to system states characterized by per-
ceived discrepancy of orientations, as distinct from sheer uncertainty as
to the other's orientations. The latter (uncertainty strain) without the
former may hypothetically occur under the following conditions: (1)
system states in which attraction is predominantly negative (dislike
and/ or mistrust); and (2) system states characterized by absence of

400 THEODORE M. NEWGOMB

object-relevance (especially those In which cathectic orientations are
regarded as matters of "taste"). The latter may be regarded, for present
purposes, as trivial. As to the former, the conditions of uncertainty strain
are presumably the same as those of discrepancy strain, as outlined in
the preceding discussion of negative strain.

Instigation to communication is a learned response to strain. This
proposition merely asserts that among socialized human beings a process
of operant conditioning has taken place whereby ( under the stated
conditions) a threatening state of affairs (i.e., phenomenal strain, either
of discrepancy or of uncertainty ) leads to instrumental behavior directed
toward the removal of that state of affairs. It is implicit in this assertion
that communication is an instrumental behavior which socialized human
beings have found rewarding because of its efficacy in removing or
reducing the threat, i.e., by establishing or increasing perceived similarity
and/or certainty. This proposition does not assert that instigation to
communication is the only learned response to phenomenal strain, but
only that in the life history of socialized humans it has been rewarded
with sufficient regularity to have been dependably learned. Neither,
of course, does the proposition say anything about the possibility that
among other learned responses to strain may be instigations to behaviors
which inhibit or prevent communicative behavior.

Instigations to initiate communication may occur under any of the
following states of system strain (in all of which one or more varieties
of positive attraction are assumed to exist, momentarily at least) :

1. The cocommunicator is perceived to "possess" an item of in-
formation which the person himself lacks and wants (e.g., "What time
is it?").

2. The cocommunicator is perceived as lacking an item of informa-
tion which the person himself "possesses35 and wants the other to have

(e.g., the information that he needs to borrow a dollar) .

3. The person wants to confirm a tentative observation (e.g.3 "Did
you hear that noise?53 ) .

All of these are instances of cognitive discrepancy, as distinguished
from the following instances of cathectic discrepancy :

4. The cocommunicator is perceived as devaluing an object which
the person himself values, or vice versa. Under these conditions com-
municative behavior (if the instigation is not inhibited) may take the
form of attempting to persuade the other to one3s own point of view
(which, if successful, would be strain reducing) . It may take the form
of "exposing53 oneself to persuasion by the other (whose success would
be strain reducing) . If there is a series of communicative exchanges, both
of these kinds of communication may occur, often with some degree of

Individual Systems of Orientation 401

resulting compromise or rapprochement — -which, again, would be strain
reducing.

5. Finally, mere uncertainty as to cathectic discrepancy may serve as
an instigator to inquiry or assertion, either of which may be instrumental
to the reduction of uncertainty.

Under certain conditions strain reduction follows communication.
This occurs most dependably, perhaps, in respect to cognitive dis-
crepancy— i.e., following the exchanging of unevaluated information.
The principal limiting conditions are those varieties of positive attraction
which have been labeled "respect53 and "trust.55 In the case of cognitive
strain reduction as a consequence of receiving information, it is at-
traction toward the other which is the limiting condition ; in the case of
transmitting information, strain reduction is most likely when the other
is perceived as respecting and trusting oneself.

The conditions of cathectic strain reduction following communication
are more complex. As everyone knows, a communicative exchange can
easily lead to increased discrepancy, and often to increased strain. The
limiting conditions here are of the following classes :

1. Conditions of cognitive strain reduction. Since cathexis pre-
supposes cognition (objects are cathected not "as they are,55 but as they
are cognized), the conditions of respect and trust, as mentioned above
in respect to cognitive discrepancy, are also operative in respect to
cathectic discrepancy.

2. Extrasystem conditions — i.e., conditions external to the system of
orientations under immediate scrutiny. Of particular importance is
"committedness" to existing attitude, which may be an aspect of a more
or less generalized personality characteristic, or which may be anchored
in other systems of orientations, e.g., in "reference groups5' of strong
positive attraction. The stronger the existing attitude committedness, the
less likely that strain reduction will occur via attitude change following
communications. The implications of this are not necessarily that the
amount of strain tolerated varies with committedness, but simply that
other modes of strain reduction are more likely to be resorted to under
conditions of strong committedness.

In general, whether with regard to cognitive or to cathectic dis-
crepancy, the conditions under which communication is most likely to
be followed by strain reduction are the same as those under which com-
munication is most likely to follow phenomenal strain, as noted in the
preceding section. This generalization follows from the assumption that,
in so far as general conditions exist under which communication leads
to strain reduction, under those conditions communicative behaviors are
learned as instrumental to strain reduction.

402 THEODORE M. NEWCOMB

Under certain conditions strain-reducing system change occurs in

the absence of overt communicative behavior. There are alternative
modes of reducing system strain. This and the preceding proposition, to-
gether, simply assert that phenomenal systems tend to shift from strain
toward "balance." Thus the questions arise: "What constitutes balance?55
"How is balance achieved by noncommunicative means?" "Under what
conditions does this occur?"

1. Balance (which is most easily defined as the absence of strain)
is most clearly present under conditions of exclusively positive attraction
and perceived similarity of orientations. The most extreme form of
strain would be found in a system including intense positive attraction
and maximum perceived discrepancy of attitude toward an object of
high relevance (e.g., the phenomenal system of either of two shipwrecked
men on a raft would be one of maximum strain if his strong attraction
to the other stems from perceived dependence upon him for safety, and
if he disagrees completely with him as to the use of their rudder, the
proper use of which he regards as essential to remaining afloat). Our
problems, however, have to do with changes of degree of strain, rather
than with the extreme points on the continuum, and so we turn to the
question of how a given degree of strain may be reduced.

2. It has already been proposed that, ceteris paribus, strain in-
creases and decreases with each of at least five orientation variables.
There are two general classes of ways in which these variables, in turn,
change: either by the receipt of information (whether via direct sensory
experience or indirectly, via communication), or by autistic operations
upon information previously received. The latter include rationaliza-
tions, memory losses, elaborations in fantasy, and other forms of
cognitive "distortion.55 It is by such processes that strain may be reduced
in the absence of communication.

The assumption that phenomenal systems tend to shift from greater
toward lesser strain applies to both classes of changes in orientation,
but nevertheless there is an important difference between the two.
Changes in orientation resulting from the acquiring of new information
are often strain increasing — i.e., upsetting to existing states of relative
balance — but autistic changes in orientation are far more rarely so. Re-
sponse to the receipt of information by "realistic5* increase in strain is
adaptive, in the primary sense of favoring a viable organism-to-environ-
ment relationship, since threatening events do occur in the environ-
ment. Autistic increases in strain may also be "realistically53 adaptive, of
course (e.g., subsequent recognition of previously unrecognized threat),
but the principal adaptive function of autistic change seems to be the
intraorganismic one of reducing strain. If so, it is presumably a con-

Individual Systems of Orientation 403

sequence of the fact that, whereas the environment is not constantly
imposing immediate demands which require strain increase, the ever
immediate demands of tension-relaxation are always operating.

3. Autistic strain reduction may be either supplementary7 to or sub-
stitutive for strain reduction following information exchange. The hypo-
thetical conditions under which it is most likely to occur are, briefly, as
follows. As a substitute for communication, the chief predisposing
conditions are absence of opportunity for communication (as determined
by extrasystem factors like physical inaccessibility) ; negative attraction
fa system variable) such that opportunities for communication with
the other person are avoided; and system states such that communication
with the other person about the system-object is avoided, or limited.
The last of these categories includes many possible combinations; for
example, communication with a positively attractive person about an
object of perceived discrepancy may be threatening to the attraction
relationship; or the balance of another of the person's systems of orienta-
tions (including the same object but a different other person) may be
threatened by "exposing" himself to influence by the cocommunicator
(the phenomenon of conflicting reference groups), so that communica-
tion with him on this topic is avoided or restricted.

As supplementary to communication, autistic strain reduction is
most likely to occur when the immediate effects of communication are
most strain inducing. Many instances of "the psychopathology of every-
day life" are illustrative — e.g., the "motivated misunderstanding" of
what another has said, or the assumption that the other has perfectly
understood the message which one intended to transmit but which one
has, in fact, transmitted with some error or ambiguity.

Alternative intrasystem changes may have equivalent effects upon
strain. This follows from the propositions according to which strain
varies as a function of several system variables. Even though redundant,
in this sense, such hypothetical substitutability merits a final note, since
it provides the basic rationale for the use of strain as a hypothetical
construct ("the little black box" ) .

Hypothetically, strain may be reduced under any of the following
conditions: (1) by reduction in the strength of attraction, (2) by re-
duction of object-relevance, (3) by reduction of perceived ("other's")
object-relevance, (4) by reduction of importance of the object of com-
munication, (5) by reduction of perceived ("others's") importance of
the object of communication, (6) by changes in cathexis or in cognitive
structuring of own attitudes, such that there is increased similarity with
the other's perceived attitudes, (7) by changes in perceived attitudes
(cathectic or cognitive) of the other, such that there is increased

404 THEODORE M. NEWCOMB

similarity with own attitudes.7 Any of these changes, hypothetically, may
occur with or without communication. And, in any given instance, once
any one of these changes has occurred with strain-reducing effects, the
probabilities that any of the others will occur are reduced.

The functional relationships among systematic variables that are
of most interest are those that contribute to strain. As already noted,
strain hypothetically increases with Increase in any of the following: (1)
object-relevance, (2) object importance, (3) strength of attraction to-
ward the cocommunicator, (4) perceived discrepancy between own at-
titude and that attributed to him, and (5) committedness (usually
stemming in part from extrasystem influences) to own existing attitude.
These functional relationships are straightforward enough, but what
about interrelationships among these five contributors to strain? And
what about their effects upon strain in those frequent instances, em-
pirically speaking, when a change in one of the five induces a change
in one or more of the others?

Let us assume that, at a given moment and with respect to a given
subject, a given cocommunicator, and a given object of communication,
there have recently been no equilibrium-disturbing events and that the
system of orientations is relatively free from strain. Increased strain
results from some psychological events (receipt of new information,
either by communication, or by direct sensory experience, or by autistic
processes) which increases one of these five (and perhaps other) system
variables. Suppose that, as a result of such an event, the individual's
positive attraction toward the cocommunicator has increased; hypo-
thetically then, other things equal, strain would be increased unless per-
ceived similarity of attitude toward relevant objects is increased.8 Or,
alternatively, suppose that a recent event has resulted in increased object-
relevance; the predicted consequence would be the same — increase in
perceived similarity of attitude. In either case, the degree of such change
resulting from increase in one variable is limited by the existing degree of
the other. That is, even a large increase in attraction will not very
dependably increase perceived similarity if object-relevance is low, nor
will a large increase in object-relevance, if attraction is weak. These
interrelationships are schematically illustrated in Fig. 3. As suggested
by the figure, these functional relationships are assumed to be monotonic
and asymptotic. These assumptions are consistent with available em-
pirical evidence, but it would be premature, on the basis of such evi-

7 With specific reference to strain of uncertainty, an eighth condition should be
added: by the receipt of further appropriate information. It is not clear, however,
that this condition is substitutable for those noted above.

8 Or, more accurately perhaps, thresholds for perceiving similarity will be
lowered.

1.

Individual Systems of Orientation
Level 3 of atfrccticn

Le/e/2

Level 1

Level 0

405

Degree of perceived discrepancy

FIG. 3. Schematic illustration of hypothesized relation-
ships among positive attraction, perceived discrepancy,
object-relevance, and strength of instigation to com-
municate. Solid lines represent greater degrees, and
broken lines lesser degrees, of object-relevance.

dence, to attempt to write equations for the assumed functions. (Object-
relevance, in particular, has been only very inadequately measured.)

As another illustration of function forms, perceived similarity of
orientation, viewed as an independent variable, is hypothetically related,
via strain, to communication (or,
more strictly, to instigation to com-
municate) : specifically, strength of
instigation increases monotonically
with strain. The functional relation-
ship may be described in a manner
analogous to that suggested in Fig.
3, as shown in Fig. 4 (in which
only positive attraction is consid-
ered, and in which the variable of
"importance" is omitted).

Again (with one exception) the
relationships are assumed to be
monotonic. There is evidence to

Level 3 of object-relevance

Level 2

Level 1

Degree of attraction

FIG. 4. Schematic illustration of hypoth-
esized relationships among perceived
similarity, object-relevance, attraction,
and importance. Solid lines represent
greater degrees, broken lines lesser de-
grees of importance.

suggest that strain increases step-
function—wise with perceived dis-
crepancy— specifically, that less
than a certain minimum of per-
ceived discrepancy induces no strain, and more than a certain maximum
leads to a reversal of attraction, and thereby to reduced strain.

As suggested by Fig. 4, the same degree of strain may be associated
with many different combinations of attraction, perceived discrepancy,

406 THEODORE M. NEWGOMB

and object-relevance. It is for this reason, of course, that the construct
"strain" seems required; i.e., it is more parsimonious to relate the count-
less possible combinations of orientation variables, as independent, to
the communication variables via the single construct of strain than to
do so for each of the possible combinations, separately. The use of the
strain construct, moreover, has the theoretical gain that significant
propositions can be generated much better with than without it.

These samples of the assumed functional relationship are representa-
tive of the others. All of them, whether monotonic or characterized by
step-function reversals, have a basis both in empirical evidence and in a
theoretical rationale. All of them are subject to correction and/or re-
finement with the gathering of further data.

ALTERNATIVE FORMULATIONS

There appear to be only three comparable formulations which have
been described in the literature, and the present one differs from each
of them primarily in attempting greater comprehensiveness, in one way
or another. Most nearly comparable in this respect is that of Romans
[15], whose chief concern is to describe group properties in terms of
hypothesized relationships among frequency of interaction, sentiment
(which in actual usage, though not by formal definition, is equated with
"liking"), activity, and (for certain purposes) group norms. Of these,
only the first two (frequency of interaction and liking) are treated as
empirical variables. At the group level, several of the propositions derived
by Homans are closely equivalent to some of those which have appeared
in these pages. But his makes no pretense to be a psychological system,
since it is concerned only with the interdependence of group properties.
There is no analysis of the intrapersonal processes by which, for ex-
ample, frequency of interaction increases with personal liking, and vice
versa. And, since attitude variables (as here defined) are not employed
at all, there is no consideration of the interdependence between attitudes
and attraction. Finally, perceived orientations of others, assumptions
about which constitute one of the foundations of the present system, are
not included at all in his system — not even, curiously enough, in his
discussion of "norms," a norm being defined as "an idea in the mind
of the members of a group . . . specifying what the members or other
men should do . . . under given circumstances" [15, p. 123],

From these differences between the two formulations I conclude that,
as might be expected in view of Homans's objectives, his system is less
capable than is the present one of accounting for such empirical data
as the following:

1. The observed covariation between attraction (including its forms
other than "liking") and perceived similarity of attitude.

Individual Systems of Orientation 407

2. The fact that communication variables® — of content as well as of
frequency — are a function, not straightforwardly of "Eking," but (a}
of Eking as interdependent with other variables, and (b) of negative as
well as positive attraction, under some circumstances.

What I find most seriously lacking in Homans's system is the absence
of a set of "system parts" which are both psychologically meaningful and
at the same time combinable into sociologically meaningful theoretical
systems. In view of my own predilections for the kind of system-theory
which makes concepts that are useful at one level of organization also
available to adjacent levels of organization, I find necessary a system-
theory which provides a psychological basis for the consensuses, both
accurate and inaccurate, which are a required condition for group life.
(Needless to say, there are many kinds of data at the group level which
Homans's formulation is far better equipped to handle than is the present
one.)

Festinger's systematic formulation [10] is strong precisely where
Homans's is weak. As indicated below, the variables of "pressure to-
ward [group] uniformity," perceived discrepancy, attraction to the
group, expected success in influencing others toward agreement with
oneself, and anchorage in other groups are hypothetically related, in
systematic ways, to the "force to communicate." The present formula-
tion differs from Festinger's (to which it owes much) primarily in that it
includes communicative behavior of all kinds — receiving as well as
sending messages, and nonpersuasive as well as persuasive messages —
and in that it includes cognitive as well as cathectic orientations
("opinion" is the only term analogous to "orientation" that appears in
his hypotheses). His system, therefore, makes no attempt to account for
either the occurrence or the consequences of nonpersuasive communica-
tion; and, since his only equivalent for the notion of "strain" is "force
to communicate," his system does not account for "autistic" accom-
modations among orientations, other than the reduction of attraction
with sufficient increase in perceived discrepancy. The two systems thus
differ in this kind of comprehensiveness but, so far as I know, there is no
instance in which opposite predictions would be derived from them.

Heider has developed a systematic formulation [13], much of which
is still unpublished, which corresponds fairly closely to the "system of
orientations" herein outlined, but which apparently does not attempt to
account for either the occurrence or the consequences of communicative
behavior. If Jordan's [18] theoretical assumptions may be taken as
representative of Heider's, at least one instance has been reported [40]

9Homans concedes that the term "communication" is the virtual equivalent
of "interaction," as he uses the latter, provided there is no assumption that com-
munication must be verbal.

408 THEODORE M. XEWCOMB

in which, different predictions having been made from Heider's and
from the present formulation, the latter would be supported.

EVIDENCE RELEVANT TO THE PRESENT FORMULATION

Nearly everyone who has concerned himself with the general prob-
lem of social influences upon attitude formation, persistence, and change
has noted, in one way or another, that for some attitudes intragroup
variance is less than intergroup variance [29]. This is particularly true
for groups characterized by face-to-face interaction [31] and for those
whose members (whether or not they interact in face-to-face manner)
have some awareness of group membership [23]. It is also particularly
true with regard to those attitude objects that are c "group-relevant," i.e.,
are of common concern to group members, and, in some sense, of dis-
tinctive concern to the members of a given group [8]. Such findings,
together with other related ones, suggest the very general conclusion that
— in certain kinds of groups, at least, and with regard to relevant ob-
jects— attitudes are formed, persist, and change not just privately (i.e.,
"between" the individual and the object of his attitude) but also inter-
personally (i.e., person-to-person influence has something to do with
person-to-object attitudes). Attitudes toward group-relevant objects
seem to be affected by some sort of intermember influence.

It has been almost as frequently observed that the same generaliza-
tion applies to group members themselves, as objects of attitudes.
Specifically, within-group variance on the part of a Polish- American
society in Detroit, for example, toward members of various ethnic and
religious groups is less than the variance of the same attitudes on the part
of the total Detroit population. This is only to say, of course, that at-
titudes toward group members vary in accordance with the same prin-
ciples as attitudes toward other kinds of objects. But data of this kind also
yield another generalization : within-group attitudes toward "own" mem-
bers tend to be more favorable than do between-group attitudes. This, of
course, is common knowledge; such findings are often mentioned as
illustrative of "ethnocentrism." Commonplace or not, the generalization,
or some variant of it, takes an important place in the development of
the present systematic formulation. Romans, in a systematic treatment
which in some respects parallels this one, goes so far as to offer this
hypothesis: "If the frequency of interaction between two or more
persons increases, the degree of their liking for one another will increase"
[15, p. 112]. His supporting data are drawn particularly from the Haw-
thorne studies [25,38].

Another common-sense observation now becomes relevant. Given
some freedom of choice, persons whose attitudes toward each other are

Individual Systems of Orientation 409

favorable tend to associate and to Interact with each other. Homans
takes note of this by adding "and vice versa" to the proposition just
quoted. These two observations, together, suggest that frequency of inter-
action and "liking" are reciprocally facilitatiue. It is possible, however,
that this circular effect is confounded by the homogeneity effects of
within-group interaction. Indeed, it seems likely that a second circular
effect parellels the first: frequency of interaction and homogeneity of at-
titude are reciprocally facilitative. This proposition, again, is supported
by a good deal of even-day observation (e.g., "birds of a feather flock
together" ) . Like the preceding one, however, it presupposes some degree
of freedom of choice.

These considerations bring to mind another set of facts. In so far as
persons associate and interact with one another selectively, on the basis
of homogeneity of attitudes, the selection must be made on the basis
of the judged or inferred attitudes of others, since attitudes, unlike
feathers, are not unambiguously displayed. Illusory judgments as well as
accurate ones may form the basis for selective association; notions like
F. H. Allports "impression of universality" [1] and Schanck's "pluralistic
ignorance" [42] were developed to account for just this phenomenon.
And so the proposition that homogeneity of attitude tends to facilitate
interaction must be paralleled by another, to the effect that perceived
homogeneity of attitude tends to facilitate interaction.10 Both proposi-
tions appear, on a common-sense basis, to be true; and (since illusions
of this kind tend to be corrected by continued interaction) the former as
well as the latter has predictive value. In any case, the facts seem to
suggest a third kind of circular effect: frequency of interaction and per-
ceived homogeneity of attitude are reciprocally facilitative.

These sets of common-sense observations may now be examined to-
gether. Interaction among group members seems to be reciprocally facili-
tative with favorable intermember attitudes, with "objective" homo-
geneity, and with perceived homogeneity of attitudes toward relevant
objects. If so, interesting questions are raised about the possibility of
interaction effects among the three kinds of attitude variables presumably
associated with behavioral interaction. May it be that any one of them
acts as a facilitating condition, or even a necessary condition, for one or
both of the others?

Most of the "evidential grounds" so far cited have not been bodies of
data gathered under controlled conditions, though such data are in fact
available. The generalizations so far presented are actuarial-empirical;

10 The latter proposition can be operationalized either for individuals (persons
tend to "choose" others whose attitudes are perceived as being like their own) or
for collectivities ("voluntary" groups are characterized by relatively frequent
judgments of intragroup homogeneity).

410 THEODORE M. NEWCOMB

that is, under actually prevalent conditions they seem to be more often
true than not. But the real problems begin with the investigation of
qualifying conditions, for which purpose data more exactingly gathered

are required. Before turning to these, it may be helpful to present a
schematic illustration of the conclusions and the questions to which the
informal evidence has so far led. Figure 5 is autobiographically accurate;
it represents a schematic summary to which I often turned at one period.

Samples of Relevant Evidence from Published Studies

The relevant evidence, from now on, has to do with the specific con-
ditions under which the three kinds of circular effects predictably occur,
and, in particular, evidence concerning relationships among the three

Objective homogeneity ? Perceived homogeneity

of attitude toward v * >• of attitude toward

relevant objects X\ i /s* relevant objects

~7~
/

Favorable attitudes
toward group members

FIG. 5. Schematic representation of probable and possible relationships
among four kinds of group variables, as inferred from informal evi-
dence. The arrows signify functional relationships for which both
empirical evidence and a theoretical rationale exist; the broken lines,
with question marks, signify hypothesized relationships.

attitudinal variables. From this point on the analysis will again be psy-
chological rather than sociological; i.e., individual rather than collective
variables will be employed. Samples of evidence relevant to all of the
lands of relationships portrayed in Fig. 5 will be presented.

The interdependence of frequency of interaction with positive at-
traction toward other persons. Perhaps the strongest evidence from which
frequency of interaction clearly emerges as the independent variable, in
this relationship, has been provided by Festinger, Schachter, and Back
[11, chap. 3]. They demonstrate that, in a university housing project
whose occupants had originally been assigned living quarters on a strictly
random basis, subsequent sociometric choices were closely related to
measures of contiguity and contact. Closely parallel findings are reported
by Deutsch and Collins [9], in a study of whites' attitudes toward
Negroes in an interracial housing project where contiguity was deter-
mined by chance.

Not many studies are to be found in which attraction appears as the
independent and frequency of interaction as the dependent variable;

Individual Systems of Orientation 411

perhaps the relationship has seemed too obvious. Newsletter, Feldstein,
and Newcomb [37, chap. 12], however, have shown that, in a summer
camp where boys were free at nearly all times to choose their associates,
observed "compresence33 varied closely with expressed personal liking.
A mean correlation of .72 is reported, for seven groups of 30 boys each;
since the reliability of the index of compresence is given as .84, it appears
that most of the variance in frequency of interaction can be accounted
for (under the conditions of this study) by personal attraction.

Available evidence does not indicate, however, that so large a pro-
portion of the variance in personal attraction can be attributed to sheer
frequency of interaction, as an independent variable. A significant pro-
portion of the variance, in the situations obtaining in the available
studies, must apparently be attributed to other factors, to a considera-
tion of which I now turn.

The interdependence of frequency of interaction and objective sim-
ilarity of attitudes. That the latter tends to increase with the former is
suggested by many studies, not all of which have attempted to control
for initial attitudes. Sims and Patrick [43], in a study of attitudes of
Northern whites toward Negroes in a Southern university, show that
their first-year students differed hardly at all from "typical" Northern
students in Northern universities, thus apparently ruling out any im-
portant influence of selection. Their third- and fourth-year students
differed hardly at all from Southern students in the same Southern uni-
versity, and mean attitudes of second-year students were exactly halfway
between those of freshmen and upperclassmen. The possibility that these
highly significant differences may be inflated by seH-elirmnation of
upperclassman with deviant attitudes is not excluded; but similar re-
sults in an otherwise different kind of study reported by Newcomb [31]
cannot be accounted for by student attrition.

The proportion of variance in similarity of attitude which can be
attributed to frequency of interaction varies enormously, almost certainly,
with a wide range of parameters. Chief among these, perhaps, is personal
attraction, which, under conditions of freedom to choose, is known to
vary closely with frequency of interaction. The evidence concerning at-
traction as related to attitude similarity is discussed below.

That existing similarity of attitudes tends to determine subsequent
frequency of interaction is more frequently attested by everyday observa-
tion than by properly controlled studies.11 It may be plausibly assumed

11 Partial analysis of a body of data recently gathered by the writer seems to
provide support for this proposition. Frequency of observed association of men
chosen as initial strangers to live in a student house under fraternity-like conditions
is significantly related, several months later, to certain preacquaintance attitudes.
(See History and Prospects of the System in Mediating Research, p. 416.)

412 THEODORE M. NEW COMB

that adequate data, if they existed, would show that some but by no
means all of the variance in selective association among persons in new
situations is determined by existing similarity of attitudes toward ob-
jects regarded as important. As noted in a preceding paragraph, a con-
siderable part of this variance must be attributable to interpersonal at-
traction, a variable which itself is closely related to objective and/or to
perceived similarity of attitude, as we shall see.

The interdependence of frequency of interaction and perceived
similarity of attitudes. If frequency of interaction is considered as the
independent variable, at least part of any consequent increase in ob-
jective similarity of attitudes must be presumed to occur via the inter-
vening variable of "perceived attitude of others.53 Such data have not
often been reported; in one of the few such studies, and perhaps the
earliest [31], it is shown that with regard to a rather wide range of pub-
lic issues upper-class students tended to view their own attitudes as more
like those of other upperclassmen than like those of freshmen ; freshmen,
similarly, tended to regard themselves as most like their own classmates.
Upperclassmen, of course had interacted much more frequently with
each other than with freshmen, and freshmen somewhat more frequently
with each other than with upperclassmen; it was the upperclassmen wrho
saw most similarity with each other and most difference with the other
group. Each group was more accurate in estimating the attitudes of its
own than of the other group. Such findings suggest that most, though
not all, of the variance in objective similarity of attitudes that is con-
tributed by frequency of interaction is attributable to variance in per-
ceived similarity.

No studies seem to have been made in which perceived attitudes of
others can unambiguously be regarded as the independent variable,
though at least one set of findings (see page 418) may be plausibly inter-
preted as showing that recent acquaintances tend to spend most time
with those perceived as agreeing with them. With regard to a more
specific kind of interaction, however — namely, persuasive communication
— the findings quite clearly show an inverse relationship between fre-
quency of interaction and perceived agreement; i.e., not surprisingly,
persuasive communications are most frequently directed to those per-
ceived as being in disagreement. But this finding, as noted below, varies
with the communicator's attraction toward the person perceived as dis-
agreeing with him.

In various ways, then, the evidence concerning circular effects be-
tween behavioral interaction and each of three attitudinal variables
raises questions about relationships among the attitudinal variables.
From the evidence already cited, to the effect that all three of them tend
to be associated with frequent interaction (under certain conditions, at

Individual Systems of Orientation 413

least }5 It is predictable that all three of them will tend to be associated
with each other. Again, samples of evidence relevant to the three lands
of relationship will be presented.

Personal attraction and objective similarity of attitude. Under condi-
tions of voluntary association, and with regard to attitudes toward rele-
vant objects, these two variables are proverbially associated. But excep-
tions, too, are proverbial; devoted spouses and best friends often disagree
about matters of great relevance. By way of documented evidence, a
study of a college community may be cited [31] in which attitudes
toward certain public issues were shown to be highly relevant, for the
community at large (though not necessarily for every pair of students).
Among those more than 1.5 standard deviations below the mean of the
total population (in the "approved" direction), friendship choices were
given and received with from two to three times the chance expectancy;
whereas among those equally extreme in the "disapproved" direction,
friendship choices were exchanged with approximately chance frequency.
For most of the latter group, as distinguished from most of the former,
the attitude objects were not very relevant. But even among the former
group, only some 25 per cent of all choices were exchanged among those
attitudinally similar, by this criterion. These findings are consistent with
others: comparatively little of the variance in observed personal attrac-
tion can be accounted for in terms of similarity of attitude toward any
single object.

Personal attraction and perceived similarity of attitude. It seems
altogether likely that much of the observed relationship between attrac-
tion and objective similarity of attitude is traceable to perceived sim-
ilarity of attitude. This presupposes, of course, some degree of accuracy
in judgments of similarity. As a matter of empirical fact, all the evidence
that I have seen indicates that, within face-to-face groups (and in many
other groups, too), two kinds of effects may be observed: "realistic"
effects, which result in fairly accurate judgments; and "autistic" effects,
as a result of which judgments are distorted by attraction — i.e., exag-
gerated estimates of similarity with others toward whom positive attrac-
tion is strong, or of discrepancy with others toward whom negative at-
traction is strong. The most striking instance of autistic effects known
to me is shown in responses to a questionnaire dealing with the then
recent dismissal of General Mac Arthur by President Truman [30]. On
this issue, 48 of 48 self-designated "pro-Truman" subjects responding to
the questionnaire attributed "pro-Truman" attitudes to "most of my
closest friends"; whereas 34 of 36 "anti-Truman" respondents similarly
attributed their own attitudes to their closest friends; 32 of 38 who were
"pro-Truman" and 13 of 27 who were "anti-Truman" attributed "anti-
Truman" attitudes to "most uninformed people." Although inaccuracies

414 THEODORE M. NEWCOMB

of estimates are almost invariably in the "autistic" direction, they are
rarely as extreme as this.

These responses, since they were obtained from subjects who had no
way of knowing the correct answers, must be interpreted as showing the
effects of attraction as the independent variable. Evidence which shows
the same relation, with perceived similarity as the independent variable,
does not abound. One of the clearest demonstrations is by Schachter
[41], who placed a confederate, instructed to express disagreement, in
each of several groups which were discussing a relevant issue. In all
groups the confederate was sociometrlcally rejected at the end of the
meeting, whereas in control groups (where the same confederates had
been instructed to express agreement) they were not rejected. It must be
stressed, however, that these findings come from a laboratory experiment
in which the subjects were strangers who had discussed only one issue.
In situations characterized by continued interaction, and by members'
familiarity with each other's personalities and with each other's attitudes
on many issues, it would be much more difficult to predict attraction
from perceived similarity alone.

There is a small literature on "reference group" influences on atti-
tudes which illumines the relationship between attraction and perceived
similarity of attitudes. If we assume that attraction to members of
one's own religious group, for example, tends to be positive, then ex-
periments by Kelley and Volkart [20] and by Charter and Newcornb
[7] may be so interpreted. Both experiments showed that Catholic stu-
dents for whom Catholic membership had just been made "salient"
made attitude responses more consistent with those presumably char-
acteristic of most Catholics than did comparable groups for whom
Catholic membership had not been made salient. (The former study
found this result for high school but not for college students; the latter
used university students only. )

Objective similarity and perceived similarity of attitude. That per-
ceived similarity, as a dependent variable, may be facilitated by objective
similarity, provided that there is opportunity for discovering that actual
similarity exists, seems obvious. Many studies, employing the procedure
of pretest, feedback concerning others' attitudes, and posttest [e.g., 17]
have shown that the obvious does in fact occur — provided the source of
information is considered trustworthy.

The relationship of perceived similarity, as an independent variable,
to objective similarity is quite different. The perception of another's atti-
tudes as like one's own is likely to increase their actual similarity only
in an indirect sense — i.e., ( 1 ) if the perceived similarity is greater than
the actual similarity; and (2) if the perceived similarity serves to in-

Individual Systems of Orientation 415

crease attraction to the other person, whose influence decreases the ini-
tial discrepancy. This effect may be forestalled, however, by the dis-
cover}' that the previously perceived similarity was illusory.

Thus the evidence, particularly as presented in several studies by
Festinger et al. [10], suggests that increase in objective similarity is
facilitated by perceived discrepancy, rather than by perceived similarity,
of attitude. This proposition, as Festinger is careful to note, presupposes
positive attraction, and evidence like Back's [4] suggests that the effects
vary directly with the degree of positive attraction.

The evidence so far presented may be summarized as follows:

1. Personal attraction, objective similarity, and perceived similarity
of attitude all vary, under partially known and partially unknown condi-
tions, with frequency of behavioral interaction.

2. Each of the three attitudinal variables covaries with each of the
others, under partially known and partially unknown conditions.

3. The covariation between each of the attitudinal variables and
interaction depends, in part at least, upon interrelationships among the
attitudinal variables themselves.

4. The interrelationships among the attitudinal variables depend, in
one way or another, upon some aspect of behavioral interaction. It is, of
course, the evidence for multiple interdependencies which suggests the
need for system analysis, as outlined in the preceding section.

The foregoing body of evidence has included very few distinctions
among the many possible variables in terms of which behavioral inter-
action might be studied; frequency of selective ("voluntary") associa-
tion, and of persuasive communication have been the principal ones so
far. Both common sense and the available empirical data (see page 411)
suggest that most of the variance in selective association, under condi-
tions of "free" choice, can be accounted for by variance in attraction.
Variance in communicative behavior, on the other hand, appears to
have much more complex determinants. Viewed as a dependent variable,
therefore, it is with regard to interaction as communication, and not
merely as association ("doing something together"), that much of the
variance is still unaccounted for. Viewed as an independent variable,
moreover, the contribution of behavioral interaction is much more prob-
lematic with respect to communication than with respect to association,
which may be regarded primarily as providing opportunity for communi-
cation to take place. The remaining samples of evidence, therefore, will
have to do with communicative forms of behavioral interaction, and in
particular with reference to the concept of "strain."

The work of Festinger and his associates [especially as summarized
in 10] provides a body of experimental evidence that is directly rele-

416 THEODORE M. NEWCOMB

vant.12 Their findings are presented with the specific proviso that com-
munications are considered to be those "which arise from pressures to-
ward uniformity in a group [in which] the communicator hopes to in-
fluence the person he addresses in such a way as to reduce the discrep-
ancy between them" [10, p. 6]. Their studies provide a good deal of
support for the following propositions (their terminology is here "trans-
lated55 into the language of the present paper) : ( 1 ) frequency of com-
munication varies directly with perceived discrepancy, with object-
relevance, with attraction, and with expected success in changing the
other's attitude; (2) attitude change following communication varies
directly with "pressure toward uniformity35 and with attraction, and in-
versely with anchorage in other person-relationships and with personality-
determined committedness to existing attitude.

A smaller body of evidence from Heider and his associates [14, 16]
suggests that "imbalance" is psychologically stressful. Specifically," Jordan
[18] has shown that the combination of "liking3' another person and of
perceiving his attitudes as divergent from one's own is "unpleasant.35

HISTORY AND PROSPECTS OF THE SYSTEM IN
MEDIATING RESEARCH

Though the present formulation has borrowed heavily from others,
together with their supporting empirical data, a considerable part of its
empirical support has been found in research (my own, or that of my
students) instigated by the demands of the developing formulation.
These investigations have provided partial, or in some cases striking,
confirmation for each of the following generalizations:

1. Following reports to subjects of the attitudes of attractive others
(sometimes groups, sometimes individuals), those attitude changes which
occur are predominantly such as to be strain reducing. Thus the greater
the experimentally induced increase in self-other discrepancy, the greater
the amount of attitude change and the more certainly change is in the
direction of reducing discrepancy [22, 26, 46].

2. Attitude change, following experimentally induced increase in
self-other discrepancy with attractive others, is less on the part of sub-
jects who do than of those who do not anchor their preexperimental
attitudes In agreement with other (extraexperimental) groups or indi-
viduals [46]. Strain is tolerated in the experimental situation because,
presumably, to reduce it by attitude change would induce still greater
strain in other, competing, systems of orientation.

12 Since the relevant findings are drawn from a large number of quite diverse
experiments, some of which have already been cited, further details are not pre-
sented here.

Individual Systems of Orientation 417

3. Inaccurate judgments of the attitudes of attractive others, whether
individuals or groups, tend to be strain reducing" — a process which may
be labeled "autistic displacement of others' attitudes toward agreement
with oneself3 [19,22].

4. Tolerance for discrepancy — i.e., maximum discrepancy beyond
which changes in the system of orientations are likely to occur — tends to
be an individual constant, and is related to personality measures pur-
porting to measure "conformity" [3].

5. Following the experimental presentation of information contra-
dicting the previous assumptions about others' attitudes, systems of
orientation are very likely to be changed if the information is accepted,
and unlikely if it is not accepted, a great majority of changes being strain
reducing [26]. The least common kind of change in this study was in
attraction ; the most common was an extension of the range of acceptable
attitude alternatives, equivalent to an extension of the area of "agree-
ment.33

6. Accuracy of judging others' attitudes varies directly with fre-
quency of communication within the middle ranges of "liking," and in
the higher ranges of "trust" [27]. Apparently the extremes of "liking"
introduce autistic distortions, and low degrees of "trust" make it diffi-
cult to evaluate information received.

7. The hypothesized effects of communication are facilitated by
similarity of the cognitive structuring of the communicators [39]. Ac-
cording to this study by Runkel, students who were cognitively "co-
linear" with instructors received significantly higher quiz grades than
others, regardless of the similarity between students' and instruc-
tors' cathectic attitudes; and among students living in the same house,
attraction tended to increase more among "colinear" pairs than among
others.

8. Finally, objective (as well as perceived) agreement tends to be
associated with both positive attraction and with frequency of com-
munication. Indices of the latter variable are derived in one case from
self-reports of communication, in a large organization [27], and in the
other [44] are inferred from the nature of membership or reference
groups (e.g., "ten best friends" vs. "rich people").

In so far as my own concerns have influenced these investigations,
they tend to have moved from demonstrations that the propositions of
central importance to the system find empirical support to inquiries con-
cerning their limiting conditions, and concerning additional parameters
that must be taken into account.

First among my present research concerns is the necessity of study-
ing change over time on the part of interacting persons. This priority
stems from my long-range objective of developing a "multiperson psy-

418 THEODORE M. NEWCOMB

chology3513 which, on the one hand, is faithful to the empirical facts
concerning both orientations and communicative behavior of individuals
and, on the other hand, can be applied to the attitudinal and behavioral
relationships among persons in such a way as to be empirically faithful
to the facts about groups. The best way of testing and extending the
propositions required for such a systematic approach is to start, de novo,
with a set of persons who have not yet developed, but predictably will
develop, the kinds of relationships characteristic of group members. My
current research, therefore, is being carried on among populations of sub-
jects recruited as complete strangers to one another and simultaneously
placed in a setting where, as a result of their joint responsibility for living
and eating arrangements over a period of several months, orientations
toward each other and toward many common objects will predictably
develop and change over time, and where certain indices of communica-
tion are available. Data have been obtained before their acquaintance,
and at weekly intervals during the entire period of their interaction. The
adequate testing of the propositions of any such theory as the present
one demands that propositions about change, as well as those about
interrelationships at a given moment in time, be put to test.

I should like to add, parenthetically, that my interest in testing the-
oretical propositions in this kind of setting stems not from any prejudice
against the "artificial" conditions of the laboratory, as distinguished from
the "real" conditions of a "natural" situation, but rather from the neces-
sity of creating the complete set of conditions necessary to test the propo-
sitions. I prefer to regard my present research setting as being, inci-
dentally, somewhat more "lifelike" than those commonly available in
short-term laboratory experiments, but as being essentially a long-term
laboratory.

At any rate, the data obtained in this setting have provided both sup-
port and discouragement for the systematic approach outlined here
Chief among the latter kind of findings is the discovery that, in this
setting, the predicted relationships among the basic variables are not in
fact invariably found for the total population of pairs of persons. For
example, it is hypothesized that, among pairs of persons, perceived agree-
ment on relevant and important issues varies with attraction. The ob-
tained correlations, though positive, are in many cases not statistically

*By this phrase I mean to imply that the search for order and regularity in the
behavior of ^ rmiltiperson systems may be quite as psychological in nature as the
search m single-person systems, providing only that psychological variables are
dealt with. Multiperson psychology is not a contradiction in terms, in the sense
that multiperson physiology would be, because psychological events on the part of
interacting persons affect each other with an intimacy and directness that physio-
logical events on the part of different persons, apart from their psychological

Individual Systems of Orientation 419

significant. What does appear, however, is an extremely close relation-
ship between these variables toward the extremes of positive attraction
and perceived agreement. This fact, in conjunction with another one —
to the effect that, after four months, estimates of the attitudes of closely
associating persons tend to become very accurate — has an interesting
consequence. Thus the highly cohesive subgroups which gradually de-
velop tend to be characterized not only by very high perceived agreement
(especially on generalized "values") but also by very high actual agree-
ment. With regard to generalized attitudes that do not change much,
the consequence is that eventual subgroup formation can be pretty well
predicted from preacquaintance agreement.14

Two considerations make it necessary to draw upon the system
variable of perceived similarity to account for such findings. ( 1 ) Even
the highest levels of attraction cannot be accounted for in terms of actual
agreement on first acquaintance, when estimates are not very accurate.
(2) At early stages of acquaintance, high estimated agreement predicts
about as well to high attraction as it does several months later. This, to-
gether with a good deal of other evidence, provides very strong support
for the following summary statement. The variable of perceived sim-
ilarity is necessary, though not sufficient, to account for those interper-
sonal relationships which distinguish highly cohesive subgroups from all
other possible subgroupings within a larger, face-to-face population.

This statement — which may not, of course, be invariably supported
by future research — seems to me to point to the need for some kind of
theory (not necessarily the present one) of systemlike structuring of in-
dividual orientations toward persons and toward common objects In the
world of the orienting person and the person oriented to. I draw this con-
clusion simply because the potency of the variable of perceived agree-
ment must itself be accounted for. This I have been unable to do, up
till now at any rate, without taking into account all of the kinds of
variables and constructs noted in this paper — including, in particular,
the construct of strain, without which the dynamic interrelationships
among the several elements in systems of orientation seem incompre-
hensible.

This has been a rough and perhaps premature attempt to systematize
the conditions which I believe it necessary to take account of if one is

14 Statements in this and the following paragraph are based upon findings from
a single, 17-man population; analysis of data from a second, similarly constituted
group is not yet complete at this writing. No attempt has been made here to
present complete results from the study here reported in part. Two very partial
reports have been published, and a complete report will eventually be issued in the
form of a monograph.

420 THEODORE M. NEWCOMB

to understand how It happens that human beings selectively assort them-
selves into the more or less enduring associations which influence so
much of their behavior. I should like to close with the reminder that,
though it has sometimes been necessary for me to describe the phenom-
ena of my concern in collective, or sociological, terms, I have tried to
deal with systems of orientation in terms of individual-psychological
variables and constructs. I shall be content if I have made it seem
plausible that something like "individual systems of orientation" do seem
to be operating in the world of persons and objects in which every social-
ized human lives. I shall not be content until systematic formulations
have been developed that are more adequate than this one.

REFERENCES

1. Allport, F. H. Social psychology. Boston: Houghton Mifflin, 1924.

2. Allport, G. W. Attitudes. In C. Murchison (Ed.), A handbook of
social psychology. Worcester, Mass.: Clark Univer. Press, 1935. Pp. 798-
844.

3. Barlow, M. F. Security and group approval as value systems related
to attitude change. Unpublished doctoral dissertation, Univer. of Michigan,
1953.

4. Back, K. W. Influence through social communication. /. abnorm.
soc. PsychoL, 1951, 46, 9-23.

5. Bogardus, E. S. Immigration and race attitudes. Boston: Heath, 1928.

6. Cartwright, D., & Harary, F. Structural balance: a generalization
from Heider's theory. Psychol Rev., 1956, 63, 277-293.

7. Charters, W. W., Jr., & Newcomb, T. M. Some attitudinal effects of
experimentally increased salience of a membership group. In G. E. Swanson,
T. M. Newcomb, & E. L. Hartley (Eds.), Readings in social psychology.
New York: Holt, 1952. Pp. 415-420.

8. Chowdhry, K., & Newcomb, T. M. The relative abilities of leaders
and non-leaders to estimate opinions of their own groups. /. abnorm. soc.
Psychol, 1952, 47, 51-57.

9. Deutsch, M., & Collins, M. E. Interracial housing: a psychological
evaluation of a social experiment. Minneapolis, Minn.: Univer. Minnesota
Press, 1951.

10. Festinger, L,, Back, K., Schachter, S., Kelley, H. H., & Thibaut, J.
Theory and experiment in social communication. Ann Arbor, Mich.: Inst.
for Social Research, Univer. of Michigan, 1950.

11. Festinger, L., Schachter, S., & Back, K. Social pressures in informal
groups. New York: Harper, 1950.

12. Gough, H. Reference handbook for the Gough adjective checklist.
Berkeley, Calif.: Univer. of Calif., Inst. Personality Assessment and Re-
search, 1955 (mimeo.) .

13. Heider, F. Attitudes and cognitive organization. /. Psychol., 1946,
21, 107-112.

Individual Systems of Orientation 421

14. Helder, F. Social perception and phenomena! causality. PsychoL
Rev., 1944, 51, 358-374.

15. Hoxnans, G. C. The human group. New York: Harcourt, Brace,
1950.

16. Horowitz, M. W.. Lyons, J.. & Perlmutter, H. V. Induction of forces
in discussion groups. Hum. Relai., 1951, 4, 57-76.

17. Hovland, C. I.. Janis. I. L., & Kelley. H. H. Communication and
persuasion. New Haven, Conn.: Yale Univer. Press, 1953.

18. Jordan, N. Behavioral forces that are a function of attitudes and of
cognitive organization. Hum. Relat.9 1951, 6, 273-288.

19. Kapos, A. Some individual and group determinants of fraternity
attitudes toward the admission of members of certain minority groups. Un-
published doctoral dissertation, Univer. of Michigan, 1953.

20. Kelley, H. H.? & Volkhart, E. H. The resistance to change of group-
anchored attitudes. Amer. soc. Rev., 1952, 17, 453^-65.

21. Krech, D., £ Crutchfield, R. S. Theory and problems of social psy-
chology. New York: McGraw-Hill, 1948.

22. Lau3 J. B. Attitude change as related to change in perception of
the group norm. Unpublished doctoral dissertation, Univer. of Michigan,
1954.

23. Lazarsfeld, P. F., Berelson, B., £ Gaudet, H. The people's choice.
New York: Duell, Sloan & Pearce, 1944.

24. Lewin, K. Field theory in social science. New York: Harper, 1951.

25. Mayo, E. Human problems of industrial civilization. Cambridge,
Mass.: Harvard Univer. Press, 1933.

26. McGrath, J. E. A study of phenomenal resolution of experimentally
induced strain in cognitive structures. Unpublished doctoral dissertation,
Univer. of Michigan, 1955.

27. Mellinger, G. D. Communication, consensus, and interpersonal at-
traction. Unpublished doctoral dissertation, Univer. of Michigan, 1955.

28. Miller, G. A. Language and communication. New York: McGraw-
Hill, 1951.

29. Murphy, G., Murphy, L. B., & Newcomb, T. M. Experimental
social psychology. New York: Harper, 1937.

30. Newcomb, T. M. An approach to the study of communicative acts.
PsychoL Rev., 1953, 60, 393-404.

31. Newcomb, T. M. Personality and social change. New York: Dryden,
1943.

32. Newcomb, T. M. The prediction of interpersonal attraction. Amer.
Psychologist, 1956, 11, 575-586.

33. Newcomb, T. M. Attitude development as a function of reference
groups: the Bennington study. In G. E. Swanson, T. M. Newcomb, £ E. L.
Hartley (Eds.), Readings in social psychology. (Rev. ed.) New York: Holt,
1952. Pp. 420-430.

34. Newcomb, T. M. The cognition of persons as cognizers. In
R. Tagiuri £ L. Petrullo (Eds.), Person perception and interpersonal be-
havior. Stanford, Calif. : Stanford Univer. Press, 1958.

422 THEODORE M. NEWCOMB

35. Newcomb, T. M., £ Svehla, G. Intra-family relationships In attitude.
Sociometry, 1937, 1, 180-205.

36. Osgood, G. E. The nature and measurement of meaning. Psychol.
Bull, 1952, 49, 197-237.

37. Newstetter, W. I., Feldstein, M. J., & Newcomb, T. M. Group
adjustment: a study in experimental sociology. Cleveland, Ohio: Western
Reserve Univer. Press, 1938.

38. Roethllsberger, F. J., & Dickson, W. J. Management and the worker.
Cambridge, Mass.: Harvard Univer. Press, 1939.

39. Runkel, P. J. Cognitive facilitation of communicative effects: an
empirical study. Unpublished doctoral dissertation, Univer. of Michigan,
1956.

40. Runkel, P. J. Equilibrium and "pleasantness" of Interpersonal situa-
tions. Hum. Relat., 1956, 9, 375-382.

41. Schachter, S. Deviation, rejection, and communication. /. abnorm.
soc. Psychol, 1951, 46, 190-207.

42. Schanck, R. L. A study of a community and its groups and Institu-
tions conceived of as behaviors of individuals. Psychol. Monogr., 1932, 43,
No. 2.

43. Suns, V. M., & Patrick, J. R. Attitude toward the Negro of northern
and southern college students. /. soc. Psychol, 1936, 7, 192-204.

44. Steiner, I. D. Some effects of perceived primary group pressures on
attitudes toward a national issue. Unpublished doctoral dissertation, Univer.
of Michigan, 1952.

45. Tagiuri, R. Relational analysis: an extension of sociometric method
with emphasis upon social perception. Sociometry, 1952, 15, 91-104.

46. White, M. S. Attitude change as related to perceived group con-
sensus. Unpublished doctoral dissertation, Univer. of Michigan, 1953.

A PRELIMINARY STATEMENT TO A THEORY

OF ATTITUDE STRUCTURE AND CHANGE1

DANIEL KATZ AND EZRA STOTLAND
University of Michigan

General Approach -(!} 424

Historical paradox 427

Outline of the Theory {2+ , 8, 9> 428

The structure of attitudes 428

Definition of attitude . 428

The Affective Component 429

The Cognitive Component 430

The Behavioral Component 431

Attitudes and Value Systems 432

Attitudes and motivation 434

Affective associations 435

The functional basis of attitudes 436

Proximal attitudes 436

Proximal attitudes and the need for understanding 438

Object-instrumental attitudes 438

Ego-instrumental attitudes 440

Ego-defensive attitudes 441

The Principle of Consistency 443

The Concept of Appropriateness 447

A typology of attitudes 449

ASective associations 450

Intellectualized attitudes 450

Action-oriented attitudes 451

Balanced attitudes 452

Ego-defensive attitudes 452

Behavior and the Expression of Attitudes 453

Assumptions about Attitude Change 456

1The authors are indebted to Irving Sarnoff and Charles G. McClintock,

their former colleagues, whose theoretical and research contributions made this
paper possible.

423

424 DANIEL KATZ AND EZRA STOTLAND

Summary 464

The Relation of the Present Formulation to the Organizational Plan of This

Volume 465

Background Factors and Orienting Attitudes {!> 466

Structure of the system: independent, intervening and dependent variables

{2+} 468

Ego-defensive attitudes toward out-groups 468

The ego-defensive attitude of conformity 469

Proximal attitude 469

Object-instrumental attitudes 469

Ego-instrumental attitudes 470

Affective associations 470

Barriers Blocking General Theoretical Advance in Psychology {12} . . . 471

References 472

GENERAL APPROACH

One general theoretical approach to the problems of psychology
zeros in on some limited aspect of behavior which occurs under restricted
conditions but which permits precision in experimentation and the use
of mathematical models. This approach can use a very simple model
of man because it is dealing only with behavior which can be manipulated
in the very limited conditions of the laboratory. Another approach to the
problems of psychology is more concerned with accounting for the wide
variance in behavior as it occurs characteristically in the real social
world. Here interest begins with an attempt to identify the significant
variables in social behavior and to understand and predict major social
outcomes.

Both approaches are probably necessary for the development of our
science, though psychology is not exempt from the impact of fad and
fashion which may give a single approach current prestige and popularity.
The merit of the first approach is its ability to produce firm knowledge
and to be relatively free of the nonscientific pressures of the practical
world. Its risk is that arbitrary and narrow limitations predispose toward
trivial outcomes. The blinders the experimenter puts on may keep him in
a scientific cul-de-sac. His limited theoretical approach may further
success in his special narrow field, but prevent his rising beyond it.
Psychophysics may develop a sound set of findings but it may never be-
come a substitute for a valid psychology of perception.

The second approach is strong in its sensitivity to the complexity
of social behavior, its possible discovery of significant variables, and
its emphasis upon major rather than minor sources of variance. It runs
the danger of overconcern with the phenotypical, of imprecision in re-

Preliminary Statement to a Theory of Attitudes 425

search, and lack of firmly based principles. In essence, the first approach
is strong in method and weak in content, the second is weak in method
but strong in content. Both approaches are needed, however; interaction
between the two can be fruitful. Laboratory experimenters can be pushed
to consider factors which seem to account for a greater share of the
variance. The social scientist can be pushed toward greater precision in
the formulation of his theory and research. Essentially we use the second
approach, but in dealing with a wide range of social facts wre are utilizing
laboratory concepts and findings in an attempt at integration. Our con-
cern is not with a model heavily restricted to one aspect of behavior but
with the attempt to apply many concepts to the social field.

Because we emphasize the second approach, we cannot follow in
detail the general plan of presentation suggested in the discussion outline
for the present study. Many circumstances enforce a more freely dis-
cursive presentation, notably, the unsettled state of the attitude area, the
incipient character of the present formulation, and (possibly) certain
limiting conditions imposed by the character of the domain under study
on the modes of systematization that may be achieved.

There is little agreement about the proper operational measures of
the important variables which should be considered and little validation
of the measures that have been used as indicators of attitudes and the
related variables of motivational processes, value systems, and defense
mechanisms. We shall attempt to define the structural characteristics of
attitudes, to describe the motivational processes related to these char-
acteristics, and to state our assumptions about attitude change, but we
cannot systematically develop the conceptual properties of all our con-
structs, point to validated operational measures for them, or describe an
appropriate mathematical model for handling the data in this field. We
regard this paper rather as a preliminary airing of considerations which
look toward theory than as a theoretical formulation. Since a position
must be developed before detailed analysis can become fruitful, at this
stage, the emphasis must necessarily be on presenting our ideas.

One difficulty to be anticipated in any attempt to achieve analytical
characterization of the present approach is that the conventional in-
dependent-intervening-dependent variable framework may not prove
entirely apt for theories which do justice to the interactional character
of social life and experience. Nearly every phenomenon of an individual's
social life which is influenced by another factor also has some influence
upon that factor. Perceptions are influenced by motivation, and in turn,
social motives are influenced by perception. Group membership can
determine attitudes, and attitudes can determine group membership. At
one stage or another in the development of an individual or of a social
unit, a factor may exert more influence than it receives. This im-

426 DANIEL KATZ AND EZRA STOTLAND

balance may, however, be mere happenstance. A general theory- should
not be invalidated because at another period the influencing factor
becomes the influenced factor. For example, during a period of social
stability, group membership may be a dominant influence on the
attitudes of members. During a period of social change, the attitudes
of people may determine group membership. Theories based upon the
proposition that attitudes are a function of group membership would be
as limited as those which hold that group membership is a function of
attitudes. To be truly general a valid psychological theory must, there-
fore, encompass both directions of influence.

This statement should not be construed to mean that some factors
do not intrinsically exert more influence than others nor that all potential
targets of influence are equally open. In a general theory, the degree
of influence attributed to various factors should not be determined by
events observed at a particular time. The degree of influence attributed
to a factor should have such sound grounding in the theory that no
shift in the interactional balance would be beyond the scope of the
theory.

To postulate that certain independent variables govern certain
dependent variables implies a noninteractional paradigm that does not
do justice to the complexities of social life. The independent and de-
pendent variables may reverse roles. For operational purposes in the
conduct of research and experimentation, the independent-dependent
formula is indispensable. It is not as useful for the development of theory.
At the theoretical level, it may be more valid to conceive of an inter-
actional system in a state of changing equilibrium but always moving
toward balance. Factors outside the system may impinge at one or more
points and the resulting change may reverberate throughout the system
until that reaches some state of balance.

We postulate that the motivational component has greatest influence
on the other parts of the system. That component is thus analogous to the
independent variable, whereas attitudes are analogous to dependent
variables. The attitude itself can be considered an independent variable
affecting behavior, however, and it can also affect the independent vari-
able of motivation. In one sense, the attitude can also be considered an
intervening variable since it is a derivative of motivation which deter-
mines behavior. Yet it violates the cardinal principle of an intervening
variable in that it has an effect upon its own independent variable. For
example, an individual's desire for an object can be influenced by
his attitude toward it. Or his attitude toward an object may determine
whether or not he experiences it as need-satisfying.

There is no logical reason why the formulation of independent-
dependent variables cannot be reversed within the conventional frame-

Preliminary Statement to a Theory of Attitudes 427

work. In practice, however, the traditional formulation tends to become
a one-way street. It may be possible to fit the interactional characteristics
of social data into the conventional framework derived from laboratory
investigations, yet one may question the advisability of forcing such an
outcome now. To be sure, such forcing may help develop sobtheories of
very limited scope, some of which can later be Integrated into a more
general theory'. But since currently so many writers are interested in
the development of limited subtheories, we have chosen to look at a wider
range of phenomena.

Historical paradox. The term attitude, If not the concept, has been
remarkably durable in the literature of social psychology. Thirty years
ago, Read Bain [9] and Percival Symonds [66] read the term out of
existence for sociologists and psychologists alike. Few have loved this
orphan child, bom In controversy and fostered in hostility, yet fewer
have been able to abandon It. It has served rather contradictory func-
tions for opposed theoretical approaches. The behavioristic system
needed the concept of attitude both for flexibility and for the oppor-
tunity of getting inside the head of the robot. The field theory of Krech
and Crutchfield needs the concept to give some stability and rigidity to
their flexible system and also to give some elements out of the total
field with which one can meaningfully work.

J. B. Watson defined the field of social psychology as the study of
attitudes [69], and the social behaviorists led by F. H. Allport em-
braced the concept to give flexibility and adequacy to the mechanistic
model of man [2]. The logic of the behaviorist system of stimulus and
response, of the conditioned response and the resulting habit patterns,
did not require an attitudinal concept. Yet dealing sensibly with human
beings, with their cognitive representations of their experiences, their
self-rehearsal of such representations, and the meaning they found in
minimal cues, did call for the additional concept of attitude. Attitude
was first introduced into the behavioristic system as a neuromuscular
set or predisposition to respond to a certain stimulus or type of stimulus
[2]. It did include verbal sets to respond, however, and soon these
verbal sets were interpreted not in stimulus-response terms but as the
subjective meaning the attitude had for the individual. Thus there was
a shift from discussing the attitude to conform to social stimulation,
which could be objectively defined, to considering radical and conserva-
tive attitudes, with radicalism-conservatism defined not objectively, as
the ends of a continuum of response, but as supporting a program with
definite social meaning. Attitude then became the back door through
which the behaviorists could be as subjective as McDougall and the
other mentalists they despised.

With the development of field theory in the hands of Krech and

428 DANIEL KATZ AND EZRA STOTLAND

Crutchfield, we have the counterpart of the behavioristic tour de force
[40]. There we have a system which emphasizes the dynamics of the
psychological field. The explanatory principles are the dynamics of the
organization of that field. The psychological processes of perception,
cognition, and motivation are interwoven into one system. In fact the
old distinction between perception and cognition is abandoned. Since
all processes are part of the same field of forces and since the forces
are always more or less in flux, we have a very fluid system. Moreover,
the determination of behavior is a matter of the organizational prop-
erties of the total field. Hence predictions postulating simple relations
between stimuli and isolated or partial processes must be abandoned.
The constancy hypothesis is rejected and the emphasis is upon the
understanding of the total field of forces.

Again, this system., which emphasizes as explanatory principles the
laws of total dynamic organization, does not logically require the con-
cept of attitude. But the practical need for taking account of behavior
does call for some stability and for some identifiable affective-cognitive
elements which can be related to social behavior and to social situa-
tions. Hence the concept of attitude is introduced to allow for the fact
that cognitive and affective organization can achieve stability and some
degree of constancy. Before the authors have concluded, however, they
are discussing elements of the cognitive structure without reference to
the total field of forces when they talk about attitudes toward racial and
ethnic groups.

One may interpret this historical paradox in two ways. First, one
may hold that a concept which can be seized upon by opposed the-
oretical systems for opposed purposes is meaningless and should be
abandoned. Or, one may contend that dealing with social realities
brought the narrowness of the systems into bold relief and that, in prac-
tice, the concept of attitude offered their extreme positions a common
meeting ground. It does not follow, but it is possible, that instead of
dropping the concept of attitude, future theoretical systems must either
include it or present similar concepts to account for essentially the same
phenomena.

OUTLINE OF THE THEORY

The structure of attitudes. Definition of attitude. An attitude can
be defined as an individual's tendency or predisposition to evaluate an
object or the symbol of that object in a certain way. Evaluation is the
attribution of qualities which can be placed along a dimension of de-
sirability-undesirability, or ccgoodness33-"badness." Evaluation in this
sense always includes cognitive and affective elements; hence presenta-

Preliminary Statement to a Theory of Attitudes 429

tion of the object or Its symbol may elicit the attitude. Judgments which
are purely cognitive would not fall into the category of attitudes. Evalua-
tions are termed high when the attributed qualities are desirable and
low when they are undesirable. A direct operational measure of an atti-
tude would be the eHcitation of verbal statements of goodness or badness
about some object or symbol; for example, the subject may check agree-
ment with the statement that Communists are untrustworthy. An indi-
rect operational measure of an attitude would be the eHcitation of affec-
tive responses such as ££I dislike foreigners." The assumption in this in-
direct measure is that the person expressing the dislike will also have a
definite evaluation of the object of his dislike.

Affective expressions do not always contain an evaluation, however.
The work of R. S. Lazarus and R. A. McCleary indicates that people
show emotional response to nonsense syllables which have previously
been associated with shock when these syllables are not consciously
recognized [43]. The concept of attitude does not include such affective
response without cognitive evaluation.

Attitudes or evaluations thus have both an affective and a cognitive
component. The amount of cognition may be minimal; it need merely
specify the object sufficiently for its recognition and relate the object to
some evaluative standard. In addition, some attitudes may have a more
elaborated cognitive component, including beliefs about the object, its
characteristics, and its relation to other objects, including the relation to
the self. Attitudes may also include a behavioral component. The be-
havioral component refers to an action tendency toward the object of the
attitude in addition to the expression of affect about it. For example, one
may regard impressionistic art as desirable but not go to a museum of
modern art, read about impressionism, or acquire prints of impressionistic
paintings. An individual who has an attitude with a behavioral com-
ponent, on the other hand, has some degree of impulsion to do something
to or about the object. In our usage, the behavioral component cor-
responds closely to the term orientation of an attitude as employed by
M. B. Smith, J. S. Bruner, and R. W. White to characterize the action
tendencies aroused by the object of an attitude [62].

The Affective Component

The affective component is the central aspect of the attitude since
it is the most closely related to the evaluation of the object. In evaluating
the object some elements of cognition are necessary; the object must
be recognized and must be related at least implicitly to other objects
and beliefs. Nevertheless, it is the affective element which differentiates
attitudinal evaluation and intellectual appraisal. A person may have
beliefs and judgments about various objects and aspects of his world,

430 DANIEL KATZ AND EZRA STOTLAND

but these are not attitudes unless an attribution of good or bad qualities
accompanies the specific belief. We believe this emphasis upon the affec-
tive component is consistent with previous theoretical and empirical
work in this field. L, L. Thurstone relied heavily upon expressions of
affect in his construction of attitude scales [67]. In applying psy-
chophysical methods to the scaling of attitudes, Thurstone and his stu-
dents first gathered statements about the issue or symbol in question
which expressed feelings of liking and disliking, of affection and of hate,
and then applied the specific scaling procedure to these statements. For
Krech and Grutchfield, the essential difference between a belief and an
attitude is that the attitude includes motivational and emotional processes
which give to the attitude its sign nature, i.e., its "pro33 or ccanti" char-
acter [4-0].

The affective loading of an attitude may vary in degree, but there
must be some minimal affect at the low end of the continuum. Attempts
to measure attitudes have frequently been directed only at the measure-
ment of the degree of affectivity of the attitude. This is the usual pro-
cedure when a rating scale is employed to measure how strongly a person
feels about the issue or symbol under study. As just noted, Thurstone
emphasized affectivity in his use of psychophysical methods in the con-
struction of attitude scales. Other workers have distinguished between a
positional dimension and an affectivity dimension. Thus, attitudinal
statements can be ordered along a continuum representing either degrees
of a logical position, e.g., conservatism-radicalism or steps toward the
accomplishment of an objective, e.g., specific actions to achieve racial
desegregation. E. A. Suchman, using a technique developed by L. Gutt-
mann, has distinguished between attitudinal position and degree of
affectivity and has obtained separate judgments on the endorsement of
statements of attitudinal position and the strength of feeling about these
statements [65]. The latter judgments give him his U-shaped curves
representing the intensity dimension on the positional scale. Nevertheless,
he sees such a close relationship between the two dimensions that he has
suggested using the lowest value of the intensity dimension to define
the zero point of his positional dimension.

The Cognitive Component

Some attitudes may be quite low in their cognitive component in
that there are few beliefs about the attitudinal object and its relations
to other parts of the world or to the individual. An individual may have
a high or low evaluation of the object but not know very much about it.
People will not only reject Turks about whom they know little but will
also respond negatively to a term like Wallonians because it must stand
for some group of foreigners [25]. The cognitive aspect can vary then

Preliminary Statement to a Theory of Attitudes 431

from knowledge of some minima! cue necessary to define the object to a
full and detailed description of the object and beliefs about it [39],
Usually these details are integrated into a logical organization of some

' O O O

degree of coherence; i.e., Negroes are a primitive, emotional, inferior
people who contribute heavily to delinquency and disease and who must
be excluded from white residential areas because property values
deteriorate when Negroes move in, etc.

The cognitive component can thus be described according to three
basic characteristics. First is the degree of differentiation., namely, the
number of cognitive elements (i.e., the number of beliefs). Second is the
degree of integration, the organization of these elements into a hier-
archical pattern. A third characteristic of cognitive structure concerns
the generality or specificity of the beliefs. An attitude with a high level
of generality includes many particular objects under the same symbol
and thus permits the same evaluation to be made in many specific
situations. A specific attitude, on the other hand, is limited to a single
object.

The Behavioral Component

Attitudes which have behavioral tendencies associated with them
are of especial interest. The individual may take steps to protect or aid
the object of his attitude; conversely, he may move to injure, punish, or
destroy the object. If he attempts to aid, the attitude is called positive.
If he tends to injure, it is called negative. This positive-negative di-
mension of behavior toward the object should not be confused with a
phenotypic description of approach and avoidance. A person may ap-
proach a prowler to do him harm and avoid interfering with a child's
interaction with his peers to aid the child's development. The effect on
the object is the criterion of positiveness and negativity. In many in-
stances, however, this dimension is correlated with approach and
avoidance. In the nature of American society, acceptance is highly
valued, whereas social rejection or avoidance usually does inflict harm
on the rejected person. The Bogardus social-distance scale is based upon
the assumption that this correlation obtains.

The cognitive and behavioral components may be closely related in
that the impulsion to action can be symbolically represented and even
rehearsed. The cognitive component also comprises knowledge of ap-
propriate and inappropriate modes of action toward the object. And, of
course, previous actions may be carried in memory. Yet these correspond-
ing cognitive elements need not be present, for an attitude to have an
action orientation; e.g., habitual motor outlets, which are the essence of
the behavioral component, may lack cognitive representation. Generally,
however, we assume presence of an action orientation if we know that the

432 DANIEL KATZ AND EZRA STOTLAND

individual has certain types of cognitive patterns. For example, if the
person has a detailed knowledge of appropriate channels of social
action, we would say that his attitude has a behavioral component. Ac-
cordingly, we would predict that the person favorably disposed toward
a political party wdll be more likely to vote if he knows where the polling
place is and when it is open, and if he believes his vote is important
for the outcome, than the person who lacks such beliefs related to an
action orientation. With respect to this problem, D. Cartwright has dis-
cussed the creation of a particular behavior structure in addition to
cognitive and motivational structures [13]. He has asserted that "the
more specifically defined the path of action to a goal the more likely it
is that the structure will gain control of behavior."

Attitudes and Value Systems

As evaluations of objects or their symbols, attitudes have a single
focalized referent. This is true even for general attitudes which may in-
clude a number of particular objects to which a single symbol applies.
This is one reason why attitudes have often been regarded in social psy-
chology as useful units for the analysis of social phenomena. Individual
attitudes, however, are frequently organized into larger structures called
value systems which are integrated about some abstractions concerning
general classes of objects. A person can have a number of attitudes, some
specific, others general, about the church of his denomination, its specific
practices of worship, its religious symbols, its specific theological doctrines.
When these attitudes are organized about some central conceptual
themes, they comprise the individual's religious value system. The term
ideology is often used to designate an integrated set of beliefs and values
which justify the position of a group or institution. But whereas ideology
is a relatively impersonal concept, the value system refers to the in-
dividual's own organization of his attitudes. Value systems resemble the
concept of sentiment as used by Shand and McDougall, save that
these authors were more concerned with the organization of different
emotional predispositions about an object or class of objects than with an
integration of different sets of evaluations [61, 47]. We use organiza-
tion and integration to denote the relating of attitudes to one another
in a hierarchical arrangement based upon abstraction and generalization.
The individual may make logical slips in the process of abstracting and
generalizing, but he does emerge with a hierarchical pattern. Later, when
new experiences show logical discrepancies in the value system, difficulty
may arise. Nevertheless, imperfect though the integration may be from
a logical standpoint, its presence means that the individual has an
organized system with some "logical" subordination and superordination
of attitudes.

Preliminary Statement to a Theory of Attitudes 433

This system is also likely to contain additional beliefs and evaluations
which justify and enrich the logical generalizations. For example, the
person whose attitudes toward big business, labor unions, and social
welfare legislation become organized into a value system of economic
radicalism may then see the leaders of big business conspiring to thwart
economic reform.

When beliefs are organized in hierarchical fashion without the in-
clusion of affective judgments, we speak of belief or cognitive systems
rather than value systems. A person may have a belief system about the
economic order which is merely an objective ordering of the facts, in-
formation, and ideas available to him.

Individual attitudes retain their identity even though they may be
part of the larger structure of the value system. And it is possible for the
same attitude to be part of more than one value system. An important
characteristic of an attitude is the degree to wrhich it is linked to a value
system. At the one extreme are attitudes which are isolated and have no
tie to a larger structure. At the other extreme are attitudes which are
thoroughly embedded in a value system. A second, though not an in-
dependent, characteristic of attitudes is the number of value systems to
which an attitude is linked. Tight linkage with one value system may
prevent an attitude from developing connections with other value
systems.

Such terms as isolation and compartmentalization are sometimes
used to refer to the lack of attitudinal integration. We shall use the
term isolation to refer to any kind of separation of the attitude from
other attitudes and value systems. We shall use compartmentalization to
refer to one type of isolation, namely, the separation of an attitude owing
to the operation of some defense mechanism. Not all isolated attitudes are
of this character. They merely may lack sufficient importance or relevance
to a value system to attain linkage. The human mind does tend to
organize its beliefs and evaluations, no doubt, but the intellectual is apt
to overestimate the degree to which the majority of people integrate
their attitudes.

Value systems are often involved in the individual's self-concept. He
has an image of himself as having certain values; hence a threat to such
a value system may arouse the same emotional response as does a more
direct threat to the ego. The extent to which a value system and its
related attitudes are tied to the self-percept is an important consideration
in attempts to modify a specific attitude [15].

The nature and extent of the tie between attitudes and value systems
has important implications not only for the amount of pressure which
needs to be mobilized to change attitudes but also for the type of force
which will produce change. A number of assumptions in this statement

434 DANIEL KATZ AND EZRA STOTLAND

need to be made explicit. First, we assume that the affective component
of the attitude will be reinforced if it is locked into a value system. The
value system will have affectivity in its own right over and above the
affective components of the individual attitudes it includes. Second, in-
formation contradicting a specific attitude which is closely linked to a
value system will result in the mobilization of some beliefs of the system.
For example, if a person who has a low evaluation of Jewrs is confronted
with evidence of their intelligence, his value system of ethnocentrism may
give him defense in depth through beliefs about the consistent political
superiority of Aryans over non- Aryans, the undesirability of miscegena-
tion, etc. Thus, the amount of informational support of an attitude, to
use a term from Smith, Bruner, and White, derives not only from its
own cognitive component but from the value system of which it is a
part [62]. Finally, the influence being brought to bear in an attempt
to change an attitude must take account both of the content of the
value system to which the attitude is tied and of the motivational force
which the value system may reflect.

Attitudes and motivation. The role of attitudes in motivating be-
havior has provoked much dispute. One school holds that attitudes help
account for the directionality of behavior but that attitudes in them-
selves are not motivational forces. For example, T. M. Newcomb re-
gards an attitude as a readiness to be motivated in a certain way but
sees the motivation as coming from other sources than the attitude it-
self [50]. The other approach views attitudes as having dynamic qualities.
Thus, G. W. Allport sees attitudes as possessing both energizing and
directional properties [5]. Our own insistence upon affect as a major
component of an attitude places us in the latter camp, since we assume
an affective process has energizing properties. Whether pleasant or un-
pleasant, affect is accompanied by neural excitation of greater intensity
or extensity than that which accompanies a cognitive process. Hence,
the person with an attitude which includes a behavioral component will
need no other motivation than the presentation of the attitudinal object
or its symbol to act positively or negatively toward the object. Yet
elicitation of the attitude does not inevitably lead to behavior toward
the appropriate object. The individual may be motivated at the moment
by stronger drives in the direction of another goal; the affective energy
of the attitude may even be discharged through the pathways serving
the ongoing behavior. For attitudes which lack an action orientation
or behavioral component, the presentation of the object will arouse the
affective process but may not lead to overt behavior toward the object.
It may lead to behavioral and verbal expressions of the emotion or to a
diffuse, general feeling of pleasantness or unpleasantness. It may express
itself by affecting other behavior occurring at the time.

Preliminary Statement to a Theory of Attitudes 435

To help discover the motivational support available to the attitude.
It is necessary to inquire into the nature and genesis of the affect which
is so central to it. Consequently, we now turn to a discussion of the rela-
tionship between basic motive patterns and attitudes. We shall distinguish
among (1) affective associations which are by-products of the process
of motive satisfaction, (2) functional attitudes in which activity directed
at the attitudinal object is satisfying in itself, and '3) attitudes which are
instrumental to the satisfaction of the other needs. In addition, attention
will be given to ego-defensive needs which involve all of the three
processes listed above. Finally, the trend toward consistency will be dis-
cussed as a motivational principle.

Affective associations. The attitudes which we term affective as-
sociations represent the spread of affect during the process of motive
satisfaction to objects which happen to be present at the time. G. Razran,
for example, has demonstrated that students showed a greater liking for
pictures after these pictures had been presented during the eating of a
meal [56]. The pleasant state aroused by the dinner had colored much of
the situation. Similarly, we may acquire unfavorable attitudes toward
many aspects of our environment because of their accidental association
with unpleasant experiences. The child who has a painful experience in
the dental chair may come away with an unfavorable attitude toward
the smell of the strong soap emanating from the dentist's hands. In other
words, the affect from motive satisfaction can spread to objects which
are not necessarily instrumental to the satisfaction of a motive but which
are associated with such satisfaction through contiguity in time and
space. These affective associations become attitudes only if the individual
thinks about them sufficiently to evaluate them. If an affective associa-
tion of this sort is not salient enough in perception or memory, it will
probably not lead to sufficient cognitive activity to become an attitude.

To be long-lasting, attitudes resulting from accidental associations
require either repetition of the original experience or a sufficiently
intense initial experience to have produced emotional arousal. The
child in the dentist's chair may not only have felt the pain of having a
tooth filled, but his sympathetic nervous system may also have been
involved, and as a result, he may have experienced fear. Had he ex-
perienced only the pain from the drilling of the tooth, he would quickly
have forgotten the incident, but since emotional arousal was also in-
volved, the experience was harder to forget. An attitude based upon such
an affective association would also be more permanent, for an ap-
propriate cue would again evoke the emotion and thus reinforce the
attitude without need for another visit to the dentist. Fears and phobias
thus can arise from a single traumatic experience. The persistence of
such attitudes has been most clearly demonstrated in the work on

436 DANIEL KATZ AND EZRA STOTLAND

avoidance learning, in which the association of some object with the
negative affect of anxiety- leads to a rearousal of the anxiety each time
the associated object is presented [48]. The persistence of attitudes based
on association between an object and a pleasant or positive affect is not as
well established as persistence based on associations with negative affect.

The functional basis of attitudes. Affective associations are not func-
tional in the individual's adjustment. In a sense, they are excess baggage
compared to attitudes which have a more instrumental function in the
satisfaction of basic motive patterns. The concept of instrumental learn-
ing has been converted into the notion of perceived instrumentality by
workers dealing with cognition, perception, and attitudes. Thus, D. Cart-
wright writes C£To induce a given action by mass persuasion, this action
must be seen by the person as a path to some goal that he has" [13].
And H. Peak has developed some of the implications of the notion of per-
ceived instrumentality for attitude change [54]. We agree with the essential
assumption that many attitudes have a functional significance for the
individual since they play a part in the satisfaction of his needs. We
do not assume, however, that the means-end character of the attitude
as a path to a goal must be perceived by the individual in the process of
attitude formation or change. Moreover, we regard this concept of
attitudes as instruments for attaining goals as so general that specification
needs to be introduced with regard both to types of motive patterns
to which attitudes are related and to the role of the attitude in motive
satisfaction. Accordingly we shall discuss three patterns which, broadly
speaking, are instrumental or functional for the individual's needs : ( 1 )
proximal attitudes, (2) object-instrumental attitudes, (3) ego-instru-
mental attitudes. The major difference among these attitudes is the
source of affect arousal. In the first case, the affect is directly as-
sociated with the object of the attitude; in the second case, it arises from
or is evoked by objects other than the object of the attitude ; in the third
case, it arises from the functioning of the ego.

Proximal attitudes. Many attitudes of the individual are evaluations
of objects which satisfy his needs and wants directly. In these instances,
the attitudinal objects are instrumental in motive satisfaction with respect
to the physiological source of the need but are consummately with re-
spect to psychological gratifications. For example, a person will place a
high value on the foods he finds especially satisfying and upon the motor
car which gives him a sense of power when he is at the wheel. These
attitudinal objects have value for the person in and for themselves and are
not easily substitutable in motive satisfaction. In contrast are the object-
instrumental attitudes in which the object is valued as a means to some
further goal. Here the object can be replaced by another object which
has the same means-value without any felt loss by the individual.

Preliminary Statement to a Theory of Attitudes 437

When we leave the area of appetitive drives, It is difficult to infer
whether the individual likes an object in and for itself or whether it is
a means to some other goal, unless we have full reports from our subject
or can study him over time. And in many instances an object can be
both satisfying for the person and also a means to other satisfactions, as
in the case of the Irishman who was paid to pull down a Protestant
church.

Since the object of a proximal attitude gives satisfaction to the
Individual when he behaves positively toward it, the attitude will receive
reinforcement with every repetition of such experience with the object.
Hence, childhood preferences for food are often difficult to change.
The mechanism by which the affect arises Is the same as in the case of
affective association, but since this latter type of association Involves
Irrelevant objects, it is not necessarily reinforced through repeated
elicitations of the same motive pattern.

Proximal attitudes are based on the principle that individuals put
high value on objects which satisfy their needs and low value on objects
which frustrate them. The need-satisfying or frustrating quality of an
object we term its functional value, yet objects which are functional from
a physiological standpoint may not be classed as "good33 or "bad.53 We do
not necessarily evaluate water or air as "good,35 although they are objects
which satisfy basic needs. An individual is most likely to evaluate func-
tional objects if he experiences some period of frustration or deprivation.
During this period, he may try various means for satisfying the need be-
fore finding the appropriate object. Easy or automatic satisfaction of
needs, as in the case of the need for air, will not lead the individual to
make an evaluation of the functional object.2

It should be pointed out that attitudes based on such direct com-
merce with the object would tend to have behavioral components, since
it is by acting on the object that the individual satisfies his motive. When
attitudes based on such functional relationships to the object have be-
havioral components, they also tend to have a well-elaborated cognitive
component. These latter aspects of the attitude often consist of the in-
formation about the object which is necessary for acting effectively to-
ward it. All the features of the object which might influence the effective-
ness of action toward it would be included in the cognitive component.

An important principle of motivation can be mentioned in con-

2 It should be understood that the distinction between readily satisfied and
frustrated needs is a matter of degree, so that attitudes can vary along a dimension
of affectivity resulting from the degree of frustration. In addition, the affective-
evaluative aspects of attitudes may not be developed in relation to experience
with the object. Some may be taken over ready-made from social groups, in
which case the degree of frustration is irrelevant.

438 DANIEL KATZ AND EZRA STOTLAND

nectlon with attitudes directed toward objects which have a functional
value for the individual. The relatively constant value of such objects
leads the individual to try to assure himself of their continuing avail-
ability. Thus his action orientation toward the object will include efforts
to protect and preserve the sources of his satisfaction. The individual will
also treat in terms of a temporal frame of reference objects with negative
functional value, i.e., those which have frustrated or harmed him. He
may seek means to destroy them or to protect himself against future
harm. Consequently, the behavioral component of this type of attitude
can include action tendencies which were not part of the original mode
of dealing with the object.

Proximal attitudes and the need for understanding. We have dis-
cussed proximal attitudes with only casual reference to the lands of
motives which lead to their formation. Although the satisfaction of
appetitive drives like hunger and sex furnishes the clearest examples of
relevant motive patterns, other motives will lead to proximal attitudes.
Among the most important needs to be mentioned in this connection
is the need to understand. Here, the motive has been variously described
as curiosity, exploration, the need to control one's world symbolically,
the search for meaning, the tendency for more inclusive and stable
organization and cognitive structure. Human beings are characteristically
troubled if they cannot obtain enough information and ideas to resolve
the confusions of a chaotic and disorganized picture of their immediate
universe. Spranger includes this pattern as one of his basic value types
[63]. Much of our everyday communication practice assumes that
supplying information to people about their problems will form and
modify their attitudes.. Much of modern communication theory im-
plicitly assumes this model, for the amount of relevant information on
the input side is an important factor in the predicted outcome.

Thus, it would be expected that those objects in the environment which
aid in understanding the world would be evaluated highly. Furthermore,
if the object itself is clearly understood, it will be evaluated more highly
than if it is understood but vaguely. The relationship between ease of
understanding an object and evaluation of it has been demonstrated
by A. R. Cohen, E. Stotland, and D. M. Wolfe, whose experiments
showed that subjects liked clearly written stories better than ambiguously
written ones [17].

Attitudes based on the need to understand will often have well-
differentiated cognitive components, since this gives the individual a
more adequate basis for understanding his world. On the other hand,
it is less likely that such attitudes will have a behavioral component.

Object-instrumental attitudes. Indirection of motive satisfaction is
characteristic of a complex society. To reach the objects we desire, we

Preliminary Statement to a Theory of Attitudes 439

must successfully negotiate lengthy and sometimes circuitous pathways.
Many objects or paths which are not rewarding in themselves can be
instruments for attaining rewards. Because of the associated affect from
goal attainment and because he perceives the instrumental value of the
intermediate object, the individual favorably evaluates the objects
which are the means of reaching Ms goals. M. Rosenberg has demon-
strated that where the individual saw the object as functional in achiev-
ing something he valued, it was regarded more favorably than objects
not regarded as instrumental [58].

Instrumental attitudes of this sort differ from affective associations
which merely receive the spread of affect from motive satisfaction. The
object-instrumental attitude, however, bears a more constant relation-
ship to the consummately satisfaction. It has been the means for reach-
ing this state and from the individual's own point of view an effective
means. It is possible to substitute another means but generally the indi-
vidual will not seek another path unless the old route was frustrating in
some of its aspects. Many of our everyday social attitudes are of this
object-instrumental character. People will have favorable attitudes to-
ward the political party which promises them prosperity. Leaders will
receive high evaluations if they can assure adequate returns to their
followers. Workers will be highly regarded by their employers if they are
high producers.

Since the indirection in reaching the goal in this pattern of motive
satisfaction occasions some delay and sometimes some degree of frustra-
tion, these instrumental attitudes usually have a considerable cognitive
component. Not only is there a recognition of the object itself but a per-
ception of the object's function as a means of reaching the goal and
frequently some notion about its appropriateness relative to alternative
means for accomplishing the same purpose. There will tend to be a be-
havioral component because the attitude is important as a route to reach-
ing some goal.

To confound our distinction between proximal and instrumental
attitudes, people in real life situations over time may find the instru-
mental object or the path to the goal rewarding in some degree in itself.
This resembles Woodworth's notion that mechanisms can become drives
[72] and G. W. Allport's doctrine of functional autonomy [4], For ex-
ample, people may find that money is such a valuable means for satis-
fying so many needs that they take pleasure in possessing and handling
it. Whether we are dealing with a proximal or instrumental attitude, the
practical test of an object's reward character is not merely the individ-
ual's report of his source of satisfaction but his behavior when the instru-
mental act is no longer fundamental for achieving the original goal.

The motivation to behave positively toward instrumental objects may

440 DANIEL KATZ AND EZRA STOTLAND

follow the principle already described in discussing proximal attitudes,
namely, the individual's need to assure himself of a continued source of
satisfaction. He may seek not only to follow the instrumental path but to
maintain it as a stable and dependable means of attaining his goals.
Should he perceive no other feasible means available for gratifying his
needs, he may evaluate the given path so highly that he tries to preserve
it. The worker to whom seniority is the only feasible path to economic
benefits will be committed to this instrumentality in a manner which his
employer may not anticipate.

Ego-instrumental attitudes. The attitudes just described are based
upon the instrumental value of the object of the attitude. Holding an
attitude, however, can have another instrumental function, namely,
maintaining the individual's conception of himself as a certain kind of
person. And in expressing such an attitude, the individual indicates to
his fellows the kind of person he is. In the proximal attitude, the affect
derives from the object of the attitude, in the object-instrumental atti-
tude, from the goal object to which the attitude is instrumental. With
the third type of attitude we are describing, the affect arises from sources
further removed from the attitude itself, from ego satisfactions. For ex-
ample, a middle-class person may hold and express attitudes which are
typical of the upper class because he sees himself as basically a member
of the upper class.

The crucial point about attitudes based on this type of motive is
their relative independence of actual interaction with the object of the
attitude. With other sorts of instrumentality, the individual has dealt
with the relevant objects and has found them satisfying either in them-
selves or because they are closely related to the end object. Thus, there
is an experiential contact with the object of the attitude that is not
essential to the ego-instrumental attitude. A person may know little about
polo, he may never have seen it played, but he evaluates it highly be-
cause holding such an attitude bolsters his self-concept.

The basic behavioral component of such an attitude is a tendency
for the person to express it to an appropriate audience as well as to
himself. He must prove to himself that he is a certain sort of person and
the expression of the proper attitudes aids him to achieve this purpose.
The cognitive component becomes elaborated less upon the basis of the
objective characteristics of the attitudinal object than upon the individ-
ual's need to maintain his own self-image.

The attitudes involved in conformity behavior may differ in nature
though their expression may not yield palpable differences to the observer
and in some cases may result in the same social consequences [8].
Some people conform to the norms of a group because they seek the
specific rewards of group membership or want to avoid group censure.

Preliminary Statement to a Theory of Attitudes 141

Their expressed attitudes are Instrumental to attaining specific objects.
Others conform because they identify with the group and see themselves
as group members; their attitudes would fall into our third category,
the ego-instrumental. In both cases, the conforming evaluations and be-
liefs are not based upon experience with the object of the attitude. On
the other hand, some evaluate the group norm in the positive fashion
that is expected of group members because the group expectations hap-
pen to coincide with their own invidual evaluations — in other wTords,
proximal attitudes based upon their own experiences with the object.
H. Kelman has suggested a similar description of conformity behavior
though his interest is not in the analysis of attitudes [35]. He distinguishes
among ( 1 ) the process of intemalization in which the individual con-
forms because the ideas and actions expected of him are intrinsically re-
warding, (2) the process of compliance in which he hopes to achieve a
favorable effect upon another person or persons, and (3) identification
in which the individual conforms because he wants to maintain a satis-
fying self-defining relationship to another person or group.

D. Riesman has described a generalized personality type which he
suggests as increasingly common in our society — the other-directed man
T57]. Such a person is sensitive to any cues which will tell him what the
group expectations are on any issue or problem. But such a generalized
trait could originate in the attempt to use the group to attain one's own
specific goals or in effort to identify with the group to support one's self-
concept. In the first case, wre have the opportunist who can exploit the
group for his own purposes. In the second case, we have the conformer
who is used by the group for its purposes.

Ego-defensive attitudes. We have separated out the four patterns of
affective associations, proximal attitudes, object-instrumental, and ego-
instrumental attitudes to describe the processes through which attitudes
develop from need gratification. Many attitudes fall clearly into but one
among these four categories. One major type, however, combines proxi-
mal and ego-instrumental functions and is, therefore, of great strength.
Ego-defensive attitudes protect the ego but their expression also gives the
individual direct satisfaction. The person who projects his own hostilities
onto other people and then attacks these hostile people satisfies two pur-
poses. Projecting his own aggression protects his self-image from a recog-
nition of undesirable qualities. Expressing the aggression gives cathartic
release. Before discussing the components of ego-defensive attitudes,
however, a more detailed account of the conception of ego defense may
be useful.

The internal conflict between our ideal self-image and impulses un-
acceptable to it evokes many mechanisms of defense. These mechanisms
seek to alleviate the conflict by keeping impulses from consciousness yet

442 DANIEL KATZ AND EZRA STOTLAND

permitting them partial expression. The classic pattern is one of repress-
ing sexual and aggressive impulses and projecting these impulses onto
others, where they can be properly attacked. The authors of The Authori-
tarian Personality have made a major contribution in showing how atti-
tudes can function in the service of these defense mechanisms [1], The
projection of our own hostility can give us gratification while maintain-
ing the fiction that these impulses originate in others.

There is an important distinction between motive patterns based
upon ego defense and those not designed to protect the self-image. In
the long view, defense mechanisms are not genuinely problem solving or
adaptive; hence, they give only partial satisfaction. The inner conflict
continues because, although defense mechanisms give some temporary1
relief, their net effect 5s to incapacitate the individual and impoverish his
emotional gratifications. The individual devotes so many of his resources
to devious means of obtaining slight satisfaction yet keeping the conflict-
ing forces apart that he operates at a low level of psychological efficiency;
in extreme instances, he will actually break down. Moreover, in his
everyday adjustment he will maintain attitudes in defense of himself
which can deprive him of rewards and incur punishment. The man who
resents his boss because he is working out some of his relations with his
father may deprive himself of advancement and the satisfaction of many
other needs. Nor will he necessarily alter his behavior because he is pun-
ished. Since he is responding to his own internal conflict, if the degree of
external reward and punishment affects him at all, it may be in the
reverse direction.

Ego-defensive attitudes thus resemble two of the motive patterns
already discussed. They are similar to ego-instrumental attitudes in that
they give the ego security through the belief that the individual is supe-
rior to others and that these other people have certain unacceptable im-
pulses. Ego-defensive attitudes are like proximal attitudes in that the re-
lease of the unconscious hostility against the attitudinal object is in itself
satisfying. All aggressive behavior does not have this satisfying quality
because the aggression can be an unpleasant means to accomplish a
desirable purpose. Not all parents enjoy inflicting physical punishment
on their children. In the case of the conflicted person, however, where
hostility has been building up, its expression gives the individual positive
satisfaction. The object of his negative attitudes may be evaluated in a
contradictory manner. The scapegoat, for instance, will be evaluated
poorly in many respects in order to justify the hostility. Thus the bigoted
person may attribute subhuman qualities to the ethnic groups against
which he directs his aggressions. Yet, he has a stake in assuring the con-
tinued availability of the scapegoat so that he may have a convenient
object for the release of his hostility. Thus, the bigoted person will often

Preliminary Statement to a Theory of Attitudes 443

protect the object from harm by others and try to keep It available to
him. A sadistic husband will not want his wife to leave Mm and may
take positive actions to assure the permanence of the marriage.

The Principle of Consistency

The motive patterns so far discussed have had to do with the
functioning of specific needs. At a more structural level, one can speak
of the principle of consistency just as gestalt psychologists talk of
principles of organization [71]. In fact, the trend toward consistency
may be the most general principle of organization of the psychological
field.

That the individual tends to make consistent with one another the
various aspects of his psychological functioning Is an old doctrine in
psychology. Freudian theory starts with the apparent contradictions in
human beha\ior and then proceeds to showr how the logic of the un-
conscious reconciles discrepancies and inconsistencies. Gestalt psychology
sees the human mind not as a collection of separate unrelated compart-
ments but as a unified organized system. Krech and Crutchfield adopt
this position when they maintain that any change in a cognitive struc-
ture will be absorbed in such a fashion as to produce minimal change
[40]. S. Asch reinterpreted the old experiments on prestige suggestion
which apparently showed inconsistency between the individual's rejecting
a statement and then, later, accepting It when it was presented as the
statement of a prestigeful person [6]. Asch maintained that this repre-
sented no contradiction on the part of the subject who made a cognitive
redefinition of the statement when it appeared in the context of a favor-
ably perceived authority and saw a different meaning in the statement.
More recently, attempts have been made to theorize about consistency
at a more elementary level. Thus Heider [26], Newcomb [51], Cart-
wright and Harary [14], and Osgood and Tannenbaum [53] have as-
sumed a tendency for the individual to achieve a state in which there
is consistency between the sign quality (either positive or negative) of
his relationship to another person and his acceptance or rejection of
the other person's communication or his attitude toward some object.

We shall assume that there is a general but limited trend toward
consistency in psychological functioning. We do not believe that the
principle of consistency is simple and sovereign. The human mind is too
complex, compartmentalization and rationalization are mechanisms too
conveniently at the disposal of human beings, and wishful thinking is too
deeply entrenched to make consistency a useful predictive tool without
detailed specifications about its operations. The pressure upon the in-
dividual to achieve consistency arises from the need to avoid conflict.
The conflict may be one in which opposing behavior tendencies are

444 DANIEL KATZ AND EZRA STOTLAND

activated at the same time, it may be one in which perceptions contradict
previous belief, or it may be one in which feelings and belief are in
opposition.

The trend toward consistency exists in its strongest form within the
confines of a single attitude; there it seeks to make the components of
the attitude congruent with one another. Inconsistencies can exist be-
tween attitudes more readily than between the components of a single
attitude.

The reason for the strength of the operation of the consistency
principle within the single attitude is that the affective, cognitive, and
behavioral components are all directly tied to the same object. Thus they
represent a molar unit of psychological functioning. It is easier to be in-
consistent in dealing with things and people when the inconsistent re-
actions are separated in time and space. If the components of an at-
titude were inconsistent so that the person would want to destroy the
thing he loved, he would be in a state of conflict. In general, then, the
consistency principle will express itself here as a tendency to achieve
a logical correspondence among the components of an attitude. The
cognitive elements will be congruent with the behavioral tendency both
in specifying the object and in describing the most effective and ap-
propriate channels of action. The expression of the behavior component
in overt behavior will in turn test out the attitudinal beliefs. The feed-
back from behavior will lead to a modification of the cognitive map to
make it a better guide to behavior in the future. The affective com-
ponents will also show a high degree of correspondence with the cognitive
component. Where the object is cordially disliked, the person will also
believe that its characteristics justify such dislike. His description of the
same object will vary from the account given by another person who
likes the object. Thus, C. Osgood and G. J. Suci found that a person's
evaluation of an object accounted for most of the variance in the at-
tribution of qualities to it [52].

Similarly, favorable affectivity will tend to be associated with positive
behavioral tendencies toward the object. This can be understood readily
by examining the conditions which led to the evaluation of the object
in the first place. Affect is related to need satisfaction. Objects which
satisfy a need or are associated with need satisfaction acquire pleasant
affect and are evaluated favorably.3 The individual will also behave
positively toward such objects, since by doing so he assumes the con-
tinuance of his present need satisfaction or of need satisfaction in the

3 In general, we can assume that positive evaluations tend to be correlated
with need satisfaction. When the need that is satisfied is to express aggression for
purposes of catharsis, however, the object of this aggression will be evaluated
poorly.

Preliminary Statement to a Theory of Attitudes 445

- * j

future. Need satisfaction leads both to positive affect and high evaluation
and to positive behavior toward the object. We assume that the In-
dividual learns that the association among the affective, evaluative, and
behavioral aspects of his attitude is the consistent and natural one. In
order to achieve consistency among the components of an attitude, he
then tends to behave positively toward objects he evaluates highly and to
evaluate highly objects toward which he behaves positively.

The concept of public and private attitudes has been introduced
into social science to call attention to apparent Inconsistencies In at-
titudes toward the same social objects. In Schanck's early study of this
problem many people In a small community exhibited one set of at-
titudes toward card playing and drinking for public scrutiny and another
set of attitudes within the privacy of their own homes r591. It is our
assumption that, although such discrepancies occur, there Is pressure
toward their reconciliation and that there will be not only rationalization
but also changes in either the private or public attitudes to make them
less discordant. In fact, Schanck's later observations of the public-private
dichotomy in the same community showed a marked change towrard an
Integration of attitudes. With the death of the community leader whose
beneficence supported the local church and with a growing perception
of the private views of others, the public attitudes changed In the di-
rection of the private attitudes. The fact that private and public situa-
tions with differing sanctions may permit initial differences In private
and public attitudes toward the same objects has important implica-
tions for social change. The propagandist constantly seeks to add the
sanction of the universality of opinion for the publicly expressed at-
titude. The social engineer may be able to employ other sanctions as
well. If people can be made to express certain attitudes in public, then,
they will be under pressure to bring their private views into line.

The principle of consistency also manifests itself strongly within the
confines of a value system. The attitudes contained within that system
will reinforce one another with respect to their cognitive, affective, and
behavioral elements. The person with a well-developed ethnocentric
value system will tend to attribute many undesirable qualities to all out-
groups, to follow discriminatory practices toward them, and to have
many beliefs justifying his evaluation and behavior. And it is more likely
that there will be discrepancies between the value systems of an indi-
vidual than to find discrepancies within a single value system. A real
estate operator can have a consistent set of buccaneering values with
respect to business practices and another internally consistent set of hu-
manitarian values with respect to the activities of his church.

Precise prediction of direction of change in reducing the incon-
sistency between components of an attitude requires knowledge of more

446 DANIEL KATZ AND EZRA STOTLAND

specific motivational forces and environmental pressures. In general,
however, we assume a priority of the affective and behavioral compo-
nents over the cognitive components. Need gratification is tied more
specificaEy to the affective and behavioral components than to the
cognitive content of the attitude. An individual may like beefsteak
and regard it as having high nutritional qualities. If he develops a
stomach disorder which makes it unpalatable, he will change his
ideas about it because the affective component of the attitude changes.
If, however, he is told by nutritional experts that steak is not especially
nourishing compared to other foods, the cognitive component of the
attitude can change but the favorable affect and the positive behavioral
approach toward it will not necessarily be modified. Cognition permits
flexibility of symbolic representation; different combinations and per-
mutations of symbols are possible.

xAltitudes with a behavioral component readily generate a cognitive
component though the reverse does not happen as frequently. After
people have bought a particular make of automobile, they find all sorts
of arguments to justify their preference. The beliefs can justify the pref-
erence and behavior and these ideas can also guide the individual's sym-
bolic behavior in regard to the object. Where the individual has sub-
stitute action tendencies he can rehearse his attitude more freely on
the symbolic level if he has some cognitive content to work with. If he
complains about something in the absence of the object, he has to verbal-
ize his complaints. Even attitudes which start out with mostly affective
elements tend to acquire more cognitive content. "I don't like that man
and I am going to find a reason" is not an uncommon phenomenon. On
the other hand, there is no similar pressure to give behavioral content
to the cognitive component. For most people it seems easier to develop
verbal elaborations and opinions than to act out their many beliefs.
The social environment, moreover, limits opportunities for action and
attaches penalties to many types of deviant action. The person cannot,
however, be imprisoned for his ideas as easily as for actions.

The behavioral component of the attitude is in general consistent
with the overt behavior which the individual directs toward the at-
tudinal object. This component derives from the original behavior in
the process of obtaining need gratification from or through the ob-
ject. On occasion, however, the individual will not respond toward an
object as one might predict from a knowledge of his action orientation.
Fear of punishment, rewards given for certain forms of behavior, and
similar factors may lead the person to behave in ways inconsistent with
his attitude. Yet over a period of time, the individual will tend to achieve
consistency between his action orientation and his expressed behavior.
He may change the behavioral component of his attitude if there is a

Preliminary Statement to a Theory of Attitudes 447

continued situational constraint against its expression; or he may strive
to avoid situations in which such constraints are present. K. Clark
has shown how prejudice will be reduced when social pressures force
the individual to behave in a nonpre judicial way [18],

Our emphasis has been upon the self-contained consistency of a
single attitude or a single value system. Compatibility and congruence
are readily found at this level. In any individual, however, we ex-
pect to find many inconsistencies among Ms different attitudes and
among Ms different belief systems. These discrepancies do not generally
create conflict for him unless the situation forces him. to make a choice.
Then he will attempt some reconciliation. The logical model of man is
too simple to do justice to the complexities of social behavior in spite
of its persuasive resurrection by the phenomenonologists.

We have omitted, however, one important source of consistency
among attitudes or among value systems — the self-concept. The self-
concept is the comprehensive value structure of the personality7. It
accounts for a major share of the congruence among value systems. The
individual sees himself as a certain kind of person and also sees himself
as holding attitudes appropriate to that kind of person. The attitudes
fit his role and status in life or they fit some role or status to which he
aspires, as in the case of nouveau riche. A stable attitude of the self gives
the individual a feeling of continuity and integrity. This is probably basic
to Festinger's concept of a drive to evaluate the self [22]. Nonetheless,
for most people the self-concept does not embrace all attitudes and
values. Moreover, some people have fluctuating concepts of themselves.
Individuals with fluctuating concepts of their traits and abilities are apt
to be more poorly adjusted and less effective group members, as Brown-
fain has demonstrated [11].

The Concept of Appropriateness

In discussing attitudes and the principle of consistency, we have
concentrated upon the structure of the attitude within the individual.
We believe that an adequate social psychology should take account of
the world outside the person, and not in the manner of the field
theorists who are concerned with environmental factors only as they
are represented in the individual's own psychological life space. Just as
a knowledge of the stimulus is important to the laboratory psychologist
working with sensory processes, so the social psychologist must con-
sider the social environment independently of the way the subject
happens to perceive it. Despite practical difficulties of time and expense,
social research ideally should have independent descriptions from a
number of trained observers of the characteristics of the social situation

448 DANIEL KATZ AND EZRA STOTLAND

in which behavior occurs and about which subjects report their per-
ceptions.

To the extent that we have knowledge of the objective situation
and of the subject's attitude, we can justifiably speak of the appropriate-
ness of the attitude. By appropriate, we mean the correspondence be-
tween the beliefs of the person about his world and the characteristics
of the world as agreed upon by independent objective observers.

If the cognitive component of an attitude does not accord with the
actual characteristics of the object of the attitude, we shall describe the
attitude as inappropriate. If beliefs about the object and its relation
to other events and about possible causes of action are erroneous, the
attitude is also inappropriate. The concept of appropriateness parallels
that of veridical perception in perception theory. Equivalent notions are
common in abnormal psychology, where we speak of hallucinations and
delusional systems. But the same unconscious mechanisms operate in
the normal person to produce distortions of and elaborate subjective
additions to the objects of the attitude.

The principle of consistency can operate to produce appropriateness
in the cognitive component of the attitude. When the cognitive com-
ponent of an attitude and the perceived characteristic of the object
appear inconsistent, the individual will tend either to change the
cognitive component or perceptually to distort the object of his at-
titude. Many studies have demonstrated the latter process [44, 21, 70],
but few have shown that attitudinal components change to be more
realistic. The reason for this imbalance probably lies in the fact that
the appropriateness of a cognitive component of an attitude is a relevant
consideration only if the object of the attitude can be perceived clearly
and unambiguously, as will be discussed later.

Stereotypes are examples of inappropriate attitudes. The person with
a stereotyped belief does not utilize the information available in the
situation but defines the situation in terms of his preconceived opinions
and reacts accordingly. This lack of discrimination of the objective
world may produce repetitive and apparently consistent behavior. It is
consistent, however, only if we center on the individual himself and
neglect the relationship between his behavior and the situation to which
it is directed. He does not take account both of internal consistency and
of the appropriateness of behavior to the environmental requirements.
Some gestalt theorists have attempted to write the notion of stereotypes
out of psychology because of the implications of blind or stupid behavior
[6]. They believe that a stereotyped attitude is a value judgment of the
observer and that all attitudes make sense to the subject in question. We
do not believe that this phenomenological question is critical, but even
from the point of view of the organization of the psychological field it

Preliminary Statement to a Theory of Attitudes 449

Is necessary to distinguish between the highly differentiated and dis-
criminating beliefs which permit true seneraiization and the simple
rigid structures which have been called stereotypes.

One precaution must be mentioned, however, in determining the
stereotypy or inappropriateness of an attitude. Appropriateness is a con-
cept useful for dealing with attitudes directed at relatively clearly de-
fined objects. In many social situations, however, the objects of attitudes,
such as other people or groups, often have characteristics or traits which
are ill-defined and ambiguous. People may behave inconsistently; a
person may behave intelligently on some occasions and stupidly on others.
If the actual characteristics of the object of an attitude are difficult to
ascertain, it is not particularly helpful to talk about the appropriateness
of the attitude, since there is no base line from which to measure the
deviation of the cognitive component of the attitude. Nevertheless, if
the object of the attitude has ascertainable characteristics which are
highly variable over time or space, any stereotype would be inappropriate.
In that case, any appropriate attitude would have to take into considera-
tion the high degree of variance.

One source of individual striving toward appropriate attitudes is
their adaptive value. In some situations, however, the maintenance of
inappropriate attitudes is adaptive for the individual. One such situation
is found when the members of the individual's group share inappropriate
attitudes. In order to communicate with them, to be accepted by them,
and to relate to them, the individual may be obliged to partake of their
inappropriate attitudes. Having this inappropriate attitude thus has
object-instrumental value for him. In this case, the individual may feel
little pressure to make his attitude appropriate.

A typology of attitudes. The following typology of attitudes is
based upon our analysis of the components of attitude structure. It
also takes into account the major distinction between ego-defensive and
other types of needs. There will be some overlap, therefore, with the
distinctions already made among motive patterns underlying the forma-
tion of attitudes. Attitudes can thus be grouped into five types: (1)
affective associations, (2) intellectuaKzed attitudes, (3) action-oriented
attitudes, (4) balanced attitudes, and (5) ego-defensive attitudes. In-
tellectualized and balanced attitudes are consistent with the older model
of a rational man who either seeks understanding of his world or
follows patterns which maximize rewards and minimize punishments.
Attitudes comprising the remaining categories deal with so-called
irrational behavior which requires an understanding of the individual's
own internal logic. These types of attitudes are similar to the empirical
categories described by Bettelheim and Janowitz in their Dynamics of
Prejudice when they distinguish among (1) the intensely anti-Semitic

450 DANIEL KATZ AND EZRA STOTLAND

veteran (action-oriented), (2) the outspokenly anti-Semitic veteran (the
affectively oriented), and (3) the stereotyped anti-Semitic veteran (the
cognitively oriented) [10]. We have elaborated our types more fully and
with more attention to their theoretical nature and basis.

Affective associations. These attitudes have minimal cognitive con-
tent and little or no action orientation. Thus, it is not possible to predict
from such attitudes to the individual's behavior toward the object itself.
In addition, this sort of attitude is not related to the individual's cognitive
structure. It stands alone and isolated from the person's belief systems. It is
an evaluation based heavily upon affect arising from the object itself.

The major source of affective attitudes is past association of the
object with need satisfaction. This association is not instrumental but
simply contiguous in time or place. Since the object is not instrumental,
its association with need satisfaction is fortuitous. Hence one cannot
predict a person's affective associations from his present motivations for
they stand apart from his present motivational structure.

As this type of attitude is essentially affective, it is logical to expect
that another of its sources should lie in the value system of the in-
dividual. The object of the attitude might be only an instance of the
class of objects of the value system. But since value systems are organized
through the interrelationship of the cognitive components of attitudes,
affective associations would tend to remain unintegrated into larger
systems.

Affective associations lack a behavioral component because they are
not intrinsic to the satisfaction of needs. To achieve such satisfaction
the individual does not have to do anything to the object except express
his affectivity.

Since affective associations have so few cognitive elements with
which contact can be made, they are difficult to change through in-
formation or verbal communication. New affective associations are
generally necessary if the old attitude is to be modified.

Intellectualized attitudes. Many attitudes have a heavy cognitive
component, in addition to their evaluative core, though they lack be-
havioral structure. Consequently they cannot be used very reliably for
predicting behavior toward the object of the attitude.

The major motivation for this type of attitude has already been
discussed. Beliefs about the object can satisfy a strong need within the
person, the need to understand the world about him realistically and
coherently. The existence of such a need has been shown by a number
of workers on curiosity needs, exploratory drives, cognitive stress, etc.
The individual's beliefs about the object can serve to satisfy this need;
hence, the individual develops a high evaluation of the object. Further-
more, as previously indicated, the individual will have low evaluations

Preliminary Statement to a Theory of Attitudes 451

of the object If Ms beliefs about it are confused and incoherent, or If
he finds his beliefs Inconsistent with his percepts of the object. When
the evaluation of attitudes based on this need to know Is high, one
might expect the beliefs to be appropriate and highly differentiated.

Such attitudes tend also to be Integrated with the cognitive systems
of the Individual. The need for a coherent view of the world evokes
effort to encompass many objects and classes of objects In a coherent
scheme, provided that facts offer some basis for such Integration. The
heavy cognitive content of these attitudes also makes possible many
areas of connectedness so that they can be organized Into a complex
value system.

These attitudes are susceptible to change through new percepts,
since they tend toward appropriateness. Since they tend toward Integra-
tion, intellectuallzed attitudes are also susceptible to influence through
changes in the cognitive structure. Should these two sources of change
clash, as they may, the individual is faced with serious conflict.

Intellectualized attitudes may arise not only from the need to under-
stand the world but from the need for self-consistency or some other ego-
Instrumental need. Attitudes derived in this fashion might tend to be
inappropriate yet have some degree of differentiation.

Action-oriented attitudes. People can satisfy their needs and develop
action tendencies toward valued objects with a minimum of cognitive
representation. This occurs when the need can be satisfied simply and
directly. Before the advent of Freudian psychology, It was assumed
that needs associated with cleanliness and body functions led to attitudes
without cognitive structure. In any event postponement, blocking, and
Indirection in the satisfaction of needs does lead to cognitive activity
and cognitive structure and so reduces the number of action-oriented
attitudes. On the other hand, the fact that social structure provides
ready-made channels for the satisfaction of many needs for many people
makes possible action-oriented attitudes. So long as there is little con-
flict, people may accept the established pathways to their goals with a
minimum of intellectualization. Surveys which investigate national
samples and explore the attitudes of people outside the academic world
find a paucity of beliefs about many problems. Nevertheless, there often
are evaluations and action tendencies toward social objects though the
cognitive exploration yields poor returns.

Action-oriented attitudes, moreover, can have a substitute activity
function rather than an orientation toward social action. They give
tension release if they are not in the service of an appetitive drive. Fre-
quently such attitudes are difficult to change because of the private
nature of their rewards for the individual. Other action-oriented at-
titudes which are more socially directed can be changed through the

452 DANIEL KATZ AND EZRA STOTLAND

development of new sources of need satisfaction, through changes In
the need-satisfying qualities of the object, and through finding new paths
toward the same goals.

Balanced attitudes. Many attitudes in the service of biological and
ego needs are fully developed, with elaborated cognitive content and
action orientation to supplement the affective core. These are the at-
titudes which have often been assumed to be the dominant orientations
of people in the economic and political realms. And in fact the platforms
of political parties are often directed at such dispositions wrhich are
supposed to characterize various interest groups or blocks of voters.

These attitudes have their source for the most part in trial-and-
error learning in achieving motive satisfactions. The process of learning
is complex enough to involve ideational processes, and the resulting
beliefs help identify pathways to the goal. Beliefs also develop to justify
the course of action. The behavioral component is necessarily built in
since the rewards are directly related to the activities pursued, so that
the arousal of the need reinstates both memories of the goal and an
impulsion toward specific action.

The belief component can be further enriched through cognitive
structures related to any aspect of this particular pattern of motive
satisfaction, including the attitudinal object. The process may be more
complex in that the evaluation of the object can be related to a specific
value system which, in turn, is connected with some cognitive structure.

Attitudes of this type tend to be differentiated in cognitive structure,
especially if the means-goal relationship exhibits complexity. Moreover,
they are generally not compartmentalized because the individual is try-
ing to maximize their satisfaction and not to protect himself against
the operation of the motive.

Attitudes in the category under discussion permit predictions of col-
lective behavior. People are not only consistent in their patterns of need
satisfaction but they are sufficiently alike in a cultural setting so that,
given knowledge of rewards and punishments in a situation, fairly
good predictions can be made. Balanced attitudes can be changed
through manipulating the external punishments and rewards, the path-
ways for avoiding and approaching these goals, and the perceptions of
such pathways.

Ego-defensive attitudes. These attitudes also have all three com-
ponents of affectivity, cognition, and action tendency in fair measure,
but they differ from balanced attitudes in the nature of the motivational
processes to which they are related. Balanced attitudes function in the
interests of more consciously recognized and acceptable needs, such as
physiological drives or the ego needs for affiliation, achievement, ex-
pression, or self-determination. Ego-defensive needs arise from internal

Preliminary Statement to a Theory of Attitudes 453

conflict and the resulting behavior is directed at an object which can-
not be an instrumental means for resolving the conflict.

There is a very tight relationship between the content of the attitude
and its motivational source in the ego-defensive attitude. The person
who expresses hostility toward those he regards as the aggressive, in-
ferior members of a minority group, and thus obtains some temporary
relief from his own inner conflicts, cannot readily meet his problem by
attitudes of acceptance and cooperation. There is little flexibility in the
types of attitudes he can hold in the sendee of this need. We cannot
necessarily say in advance whether this motivational pattern will be
directed more intensely against one minority group than another, but we
know that the probabilities are high that, outside the majority group,
it will find a scapegoat which permits the safe expression of hostility.

The relationship between personality type and social attitudes is an
old problem in social psychology. Some of the earliest work on attitude
research started with the premise that social attitudes were closely tied
to personality needs and conflicts [3]. Early attempts to show the de-
fensive nature of radicalism were not productive because they did not
choose an area in which there was any necessary connection between
a deep-lying need of the individual and attitudinal expression.

Workers in the field of authoritarianism, on the other hand, selected
a motivational pattern and its resultant attitudes where the nature of
the motive permits little flexibility in expression [1]. Though they have
overgeneralized their findings and are dealing with only one type of
attitude, their contribution is still of major proportions. Yet though we
can predict attitudes of prejudice toward minority groups from the per-
sonality syndrome of repression and projectivity, we cannot predict from
the attitude to its motivational source with the same degree of success.
Expressions of hostility toward minority groups may have other causes
than repression and projectivity. Sufficient allowance for the one-way
directionality of prediction between motive and attitude is necessary in
personality theories of attitude determination.

The cognitive content of ego-defensive attitudes can be partly
changed by external pressures; i.e., the individual may be taught not
to express his hostility against one type of scapegoat. But since the basic
motivation remains, he may seek a new type of scapegoat. Fundamental
change in ego-defensive attitudes calls for some degree of personality
change through the development of self-insight by the individual,

Behavior and the Expression of Attitudes

Researchers in the area of attitudes have often been disturbed and
discouraged by their inability to predict the behavior of an individual
from a knowledge of his verbalized expression. The public opinion

454 DANIEL KATZ AND EZRA STOTLAND

pollsters with many years of experience predicting election results still
cannot predict which of their respondents will actually vote on election
day. It is necessary to identify several sources of difficulty in dealing
with the relationships between attitudes and behavior. First, insufficient
attention has been given to the differences among types of attitude.
Attitudes which have little or no action orientation are not necessarily
good predictors of behavior. Often the investigator assumes that the
individual who expresses an evaluation of an object is also committing
himself to a corresponding form of behavior toward it.

Secondly, failure to inquire carefully into action orientations in the
measurement of attitudes can lead to incorrect inferences about the pre-
disposition to behave in certain ways. The subject may have an action
orientation but it may not be the one the investigator assumes, since
the investigator is basing his judgment on the subject's expressed liking
or disliking of some object.

Examples of this mistake in the identification of the presence or
absence of action structure in an attitude, and the precise nature of such
a structure where it does exist, can be found in many practical situations.
Social and industrial organizations expend considerable effort to create
favorable evaluations of the group and its goals but give little attention
to the action orientation of the attitude. Katz and Kahn have shown the
absence of high positive relationship between morale and productivity
in industrial organizations and have suggested two reasons for this find-
ing [33]. (a) Some workers who like the company and their jobs may
have no action orientation accompanying this favorable evaluation, (b)
Other workers have an action orientation but it is directed solely at
staying within the system. Accordingly, they may work hard enough to
avoid being fired but not as hard as their abilities would permit. Even
where favorable attitudes toward the system exist, membership in a
group or social system does not give sufficient evidence to predict be-
havior beyond minimal role requirements. Moreover, there may be more
than one set of group standards within a social system [60].

Thirdly, predicting behavior from a knowledge of single attitudes is
difficult because the same object may be tied to more than one attitude.
For example, a real estate agent may have an unfavorable attitude, with
an action orientation of avoidance, toward members of minority groups.
He also has a positive attitude toward clients who are in the market for
new homes. He may, then, encounter a minority group member who
wants to buy a new house. The choice of the attitude which is expressed
is a function of (1) the strength of the two attitudes, (2) the strength
of the present motivational forces, and (3) the context in which the
object of the attitude is perceived. The strength of the attitude refers to
its degree of affectivity as this relates to its action orientation. The

Preliminary Statement to a Theory of Attitudes 455

strength of present motivational forces refers to the needs which happen
to be uppermost in the person at the given time. The real estate sales-
man who has not sold a house in weeks is in a motivational state in
which the attitude toward a client Is more easily elicited. The context
for the perception of an attitude would Include the en\ironmental situa-
tion as well as the existing cognitive frame of reference. The minority
group member may appear In the real estate office with Influential
friends from the majority group and so may be more easily perceived as
a client Immediately preceding experience may have set up an ex-
pectation for certain types of objects and thereby may determine how
a given object will be perceived.

Fourthly, difficulties In predicting to behavior from attitudes arise
from the distinction between the object and Its symbol. The object and
Its symbol may be related In any one of several ways. An individual may
not distinguish between the object and Its symbol; In this Instance, the
prediction to behavior is made easier. The individual reacts In the same
manner to both object and symbol, as In the case of the person who
states on the Bogardus social-distance scale that he will not admit
Negroes to his club as members and acts accordingly. Or the man
grumbling to himself or his family may be rehearsing the sort of
activity that may later be directed toward the object Itself. On the
other hand, the individual may distinguish sharply between the object
and its symbol, and the attitudes toward the two may not be identical.
In many cases, moreover, social objects are not readily available as
attitudinal targets, or if they are available, behavior directed toward
them requires efforts in a social world which already overtaxes the
energy of the individual. Furthermore, in some instances, expressed be-
havior toward the social object may lead to some risk of punishment.
It is also true that people of a quietistic frame of mind may consistently
avoid involvement in the arena of practical problems, whereas the firing-
line type of person is consistently action-oriented.

The expression of an attitude may be a substitute activity which gives
some release and may actually mean less likelihood of subsequent action
against the object itself. Or, an individual may react to the symbol In
a manner consistent with his ego ideal in order to protect himself against
the devastating effects of his full awareness of his behavior toward the
actual object. The situations in which he reacts to symbols and to the
object may be so different as to elicit entirely different motivational
systems, and therefore, different attitudes.

The individual may, however, express his attitude toward the symbol
of an object in order indirectly to affect the object. The individual may
communicate his feelings to other people so that they may be more
favorably or unfavorably disposed to the object. These people as a

456 DANIEL KATZ AND EZRA STOTLAND

group may be powerful enough to behave to the object in ways that
the individual feels helpless to undertake.

Assumptions about Attitude Change

We have examined the nature of attitudes from three points of view :
(1) their structural components and characteristics, (2) their relation-
ship to the more comprehensive structures of value and belief systems,
and (3) their functional relationship to motive patterns. In this process
a number of assumptions about attitude change were suggested. We
should nowT like to make these assumptions explicit and to add further
assumptions that foUowT logically from the characteristics and functions
of attitudes which we have described.

1. The most basic assumption is that the key factors in attitude
change are not the situational forces or the amount and types of informa-
tion to which the individual is exposed but the relation of these factors
to the individual's motive patterns. The following assumptions are either
supplementary to this first assumption or detailed elaborations of it.

2. Major motive patterns may be ranked in terms of their relative
urgency: (a) biological or appetitive drives, (b) ego motives (social
drives) and ego defense, (c) curiosity, the need to understand. As drives
at one level in this order are satiated, drives at the next level become
all-important. In prisoner of war camps where hunger begins to operate
as a drive, the more complex motives lose importance. The culture
of the camp becomes a food culture, and self-preservation in its most
elementary forms becomes manifest. The need for understanding and
knowledge comes into full play only when more basic motives have been
satisfied. A well-constructed ideology can crumble overnight if more
basic patterns which it assumes are frustrated.

3. The principle of consistency is almost always operative, but
the direction it takes and the devices used to achieve consistency are
subject to definite limitations. Thus, individuals can reduce inconsistency
by giving up one of two opposed sources of need gratification. The
common solution, however, will be to keep both incompatible desires and
to resort to compartmentalization and rationalization of single attitudes
to avoid the inconsistency.

People will attempt to maintain both discriminatory practices and
democratic attitudes by rationalizing the discrimination. In one study
of this problem in college fraternities, students' reasons for barring
certain people from their fraternities were examined [32]. Of the five
reasons advanced, two were frank statements of personal objections to
the excluded groups, two concerned fear of public opinion, and one
included both personal objection and fear of public opinion. The
majority of the students, however, did not espouse the first two reasons

Preliminary Statement to a Theory of Attitudes 457

and professed no personal objection In spite of the fact that the ques-
tionnaire was anonymous. If these results are taken at face value, we
find that the majority fear public opinion althoegh they themselves are
the major portion of the public. Possibly, a condition of pluralistic
Ignorance did prevail but it Is also possible that Individuals rationalized
their own objections as the common prejudice of others.

4. Where the inconsistency does not lie deep enough to require
repression and cornpartrnentalization, the Individual will try to resolve
it by compromise rather than by rejecting one of the logically opposed
alternatives.

In the interesting experiment of Janls, Lumsdaine, and Gladstone,
the effects of a preparatory communication upon reactions to a sub-
sequent event were tested [31], The communication was a statement
about the poor prospects Russia had of developing atomic weapons;
this Information was later contradicted by the news that Russia had
exploded an atomic bomb. The experimenters wanted to find out
which of two theories would better account for the reaction of subjects
to the information that Russia had the atomic bomb. One hypothesis
predicted over-compensation, or a boomerang effect of the contradictory
news, so that people would reject completely the information from the
early communication. The other hypothesis predicted a positive effect:
the early communication would prevent complete acceptance of the
meaning of the actual event. The results confirmed the second hypothesis.
After President Truman's announcement that Russia had exploded an
atom bomb, the experimental subjects who had received the earlier
optimistic communication showed greater resistance to the impact of
this pessimistic event than did the control subjects who had not received
the earlier information. Apparently, people did not accept the literal
logical meaning of the two opposed sets of information but found a
compromise position between them.

The boomerang hypothesis does work on occasion, and we do reject
one source of information completely, but the academician tends to give
the boomerang hypothesis greater weight than it possesses. The advertiser
and the propagandist would not be flourishing today if the boomerang
effects of false claims were the rule rather than the exception.

Another experiment which shows the tendency toward cognitive
compromise is Stouffer's analysis of conflicting social norms [64]. Stu-
dents were presented with a hypothetical situation in which a proctor
found his roommate cheating on an examination. They were asked to
indicate what course of conduct by the proctor would be approved by
the other students. They also were asked to indicate the actions the
authorities would approve in the same situation. The over-all percentages
showed a conflicting emphasis of the expectations of authorities and of

458 DANIEL KATZ AND EZRA STOTLAXD

students, but the largest single group of students was able to find some
compromise by indicating at least one action in this situation which
would be approved by both authorities and students. Stouffer concludes
that a social norm is not a point but a band of permissible behavior
wrhich allows for some slippage from the logically prescribed role.

5. Attitudes wrhich wre have termed affective, action-oriented, in-
iellectualized, and balanced can all be changed through gaining control
of the individual's beha\ior towrard the attitudinal object. Organized
groups characteristically emphasize the control of behavior and make
certain actions an essential requirement for all group members. The
priority of cognitive and perceptual factors in modern psychological
theory has obscured the importance of required behavior as a determin-
ant of the individual's beliefs. Role playing, though weaker as an in-
fluence than the assumption of a role in social life, is effective partly
because it calls for behavioral change. Culbertson found that more favor-
able attitudes toward Negroes resulted from the assignment of roles in
a role-playing situation in which subjects worked out problems of Negro
housing [20]. Similarly Janis and King found that the task of making
speeches led to attitude changes which conformed to the position out-
lined in the speech [30],

In a more natural setting, Harding and Hogrefe showed that white
department-store clerks who had to work with Negroes tended to be
more favorable to this association after the fact [23]. One of the most
frequent and revealing reasons given by the clerks for their changed
attitudes was simply that they had to work with Negroes. Thus they
made their attitudes consistent with their behavior.

Attitudinal change resulting from role change was dramatically
demonstrated in the Lieberman study of workers promoted to positions
of foremen and elected to the office of union steward [45]. Both fore-
men and stewards had been included in a survey of all rank-and-file
workers before their promotions to these roles. In this first measurement
the foremen-to-be and the stewards-to-be were more critical of the com-
pany than the other workers, but the former two groups resembled each
other in attitudes and personal qualities. After a year in their new roles,
however, they differed strikingly in their attitudes. The new foremen
had taken over management ideology; the new stewards had taken over
union ideology.

6. Attitudes of the intellectualized and balanced type can be changed
through a change of the value system in which they are integrated. The
object of the attitude is but one instance of the general class of objects
to which the belief and value system refers and thus the attitude tends
to be made consistent with the larger structure. If this larger structure
or value system can be changed, the individual will tend to shift his

Preliminary Statement to a Theory of Attitudes 459

attitude accordingly. Changes in value systems may occur through
changes In the component attitudes over time or through a modification
of the generalized values brought about by some radical experience in
a drastically changed environment, A dramatic Illustration of this second
type of change Is illustrated in the experiment of Ashley, Harper, and
Runyon in a replication of the Bruner-Goodman study on the rela-
tion between, need and the estimation of the size of coins [7"1. The ex-
perimenters hypnotized their subjects after first obtaining their judg-
ments of the size of coins. \Vhile in the trance state, some subjects were
given the suggestion that they were very poor and other subjects that
they were very rich. The estimates that were then made of the coins
shifted upward for the subjects who had been told they were poor and
downward for those who had been told they were rich. Thus the change
in the value system involving the self-concept (whether he saw himself
as wealthy or poor) led to a change in the cognitive components of his
attitude toward monetary objects.

7. Intellectualized and balanced attitudes can be changed through
modifying the cognitive component of the attitude. Though motives and
behavior are assumed to be more significant determinants of psycho-
logical functioning than beliefs, the cognitive part of the attitude has
some importance in its own right. It defines the object of the attitude;
beliefs about the object though they may be rationalizations are none-
theless helpful in guiding behavior. Individuals characteristically seek
rationalizations and often must find them before they act or before they
feel comfortable about their actions. Timing is important since the in-
genuity of the individual is limited and finding the proper rationalization
takes time. With a long time span, as in an election campaign, the indi-
vidual can find the rationalizations he needs to justify his voting be-
havior. This is one reason why. In spite of fluctuations during the cam-
paign, the overwhelming majority of people show the same political pref-
erences in November that they showed in June.

The role of beliefs in behavior has been clearly demonstrated by
Raven's work on pressures on deviates to conform to a group norm [55].
Groups of subjects read a juvenile delinquency case and gave their opin-
ion about the extent to which the delinquent was personally responsible
for his crimes. Most subjects took a favorable position toward the delin-
quent. A false consensus of group norm was then reported to subjects.
They were asked to restate their evaluations and then to write a descrip-
tion of the delinquent. At a later point they were asked to give their
evaluations of the case once more. The subjects who shifted toward the
unfavorable group norm were those who gave more unfavorable de-
scriptions of the delinquent. In other words, the subjects who felt pres-
sure to conform to the group norm did not do so until they had devel-

460 DANIEL KATZ AND EZRA STOTLAND

oped beliefs about the case which supported their shift toward the norm's
attitude. A related finding occurred in Crockett's study of the use of
group norms in producing change [19]. When subjects were given new
group norms with reasons for the group position, they changed sig-
nificantly more than subjects who were presented with only the group
norms.

8. Attitudes can be changed through a modification of their affec-
tivity as a result of emotional conditioning. The discussion of affective
associations referred to the fact that to be heavily laden with feeling an
object did not need to have an instrumental value for the person. The
Razran study has been cited as an example of how such associations can
occur between satisfying a hunger need and positive evaluations of asso-
ciated pictures or musical selections [56]. Another instance is Murray's
demonstration that frightening children leads them to make low evalua-
tions of faces presented to them at the time of their fear [49].

A more complex case of the influence of associating fear with objects
is shown by the work of Janis and Feshbach on the influence of fear-
arousing communications upon the degree of conformity to the objectives
of the communication [29]. Three groups of subjects were informed
about the consequences of failure to practice proper dental hygiene. The
first group was shown pictorial material of a frightening character; the
second group saw less dramatic material; the third group received sim-
ilar information but of a nonemotional character. The three groups
were given the same instructions about care of the teeth. The follow-up
study indicated that the group subjected to the minimal fear appeal had
the most members observing the instructions. The group exposed to the
maximum fear appeal had fewest members following the suggestions
about good dental care. Apparently the emotional arousal was so strong
that it colored the situation and led to an avoidance of the problem.
Instead of the negative feelings becoming attached to improper practices
with respect to tooth decay, the unfavorable affect spread to the proper
practices. Thus, in order to predict the effectiveness of an emotional
appeal, it is necessary to know whether the affect can be narrowed to
the appropriate object. When punished severely, children may react
negatively to the punishing parent rather than to the forbidden activity.
Factors which determine the outcome are the clearness of the pattern of
the desired activity, the degree of emotional arousal, and the other asso-
ciations with the emotional source.

9. IntellectuaHzed and balanced attitudes can be changed if the
instrumentality of their objects for achieving the individual's goal can be
changed. The low evaluation of an object can be shifted upward if the
object can now be seen as a means for attaining some desirable end. This
is not a matter of developing new needs or values but rather of estab-

Preliminary Statement to a Theory of Attitudes 461

lishing new connections between behavior and existing needs. The base-
ball magnate who previously has had discriminator}' attitudes toward
Negroes can be made to see them as desirable people since they can help
him win pennants and draw crowds to the ball park. Carlson has been
successful in changing attitudes toward a minority group by making his
subjects aware of motivational consequences which they had not pre-
viously related to favorable behavior toward this group [12"!.

Experimental attempts to compare emotional with logical appeals
have not taken sufficient account of the Instrumentality character of the
appeals employed. Hartxnann reported that leaflets with an emotional
appeal were more effective than leaflets with a logical appeal in a politi-
cal campaign [24]. In his emotional appeal, however., the instrumental
value of voting in the advocated way was clearly stated. In the logical
appeal, this instrumental value was not so obviously expounded. Thus,
part of the greater efficacy of the emotional appeal may have been the
result of the exposition of the instrumental value of the attitude rather
than of its "emotionality.53 The failure of Knower and others to find any
differences between an emotional and nonemotlonal appeal may be
owing to a lack of difference in the degree of instrumentality of the ap-
peals which were used [37, 38].

Making an object instrumental for some goal or value of the individ-
ual generally requires some consideration of the frame of reference in
which the object is perceived. The frame of reference consists of the
standards of judgment at any point in time and may reflect past learning
or events just prior to the present, as well as the Immediate situational
forces. Apparent inconsistency in behavior may result from the same ob-
ject being judged at various times in different frames of reference and
therefore having different attitudes directed toward it. R. T. LaPiere
found that restaurant and hotel owners accommodated a Chinese couple
when they were approached by the couple in person [42]. Nevertheless,
they responded to a questionnaire by the flat statement that they would
not accept Chinese. The same discrepancy is reported by Kutner,
Wilkins, and Yarrow who found restaurant owners admitting Negroes
but refusing to make reservations for them [41]. The written commitment
may be of some importance here but it also is probable that a customer
seeking food or shelter, even though of a different shade of skin color,
is perceived in a different frame than a written request from a Negro.

To change the instrumentality of the object it is necessary to elicit a
frame of reference involving a positive and fairly powerful value system.
When this value system is salient in the individual's thinking, the old
attitudinal object can be introduced and its usefulness for the individual's
goal considered. Effort to change a person with a negative evaluation of
an object is peculiarly difficult, since mere mention of the object calls

462 DANIEL KATZ AND EZRA STOTLAND

forth the negative attitude and interferes with further consideration of
the matter, A frame of reference not involving the object directly, how-
ever, can be employed. Within this changed focus, the touchy object
can be introduced and connected with some important value of the
person.

10. The lasting effects of attempts to change intellectualized and bal-
anced attitudes are related to the figural or background character of the
various factors in the situation productive of change. Memory operates
differentially with respect to information received, with greater retention
and effect of figural than of background items. The need to know works
through the imperfect mechanism of retentive assimilation. Central ele-
ments from a communication may be recalled and may affect attitudes
much more than peripheral items concerning the source of the informa-
tion, the time and the place, the people present, etc. We remember events
but not their exact dates; we can quote lines but we cannot cite their
chapter and verse; we remember ideas but not their source. Hence in-
formation which is received and discounted at the time, because of the
attendant source and circumstances, will later have its effect upon the
individual. This fact has been exploited by propagandists with the tech-
nique of the big lie, the repeated lie, or the whispering campaign.

A number of experimental findings can be explained on this basis.
Hovland and Weiss presented the same communications to matched
groups of subjects [28]. In the one instance, the communication came
from a highly credible source; in the other case, from an untrustworthy
source. The immediate effects upon attitudes were in the expected direc-
tion, but four weeks later the results from both sources were about equal.
The sources of the communication had been forgotten, as far as any ap-
preciable effect upon the message was concerned. Kelman in another
experiment reinstated the sources in an otherwise replicated procedure
and with the reinstatement came the original effect [36].

The "sleeper effect" in which the influence of the communication is
greater over time than shortly after the reception was reported by Hov-
land, Lumsdaine, and Sheffield [27]. Short-time and long-time effects of
a film "The Battle of Britain" designed to strengthen confidence in
America's ally, Britain, were measured in two matched groups of Amer-
ican soldiers. The group tested immediately after the film showed greater
memory for factual content of the film, but the group tested after nine
weeks showed a greater change in favorable attitudes toward Britain's
role in the war. Thus when the situational pressures against accepting
the film were removed or forgotten, the changed beliefs about the object
produced changed evaluations of it.

One of the hypotheses advanced by the experimenters is the same as
the assumption we have made, "forgetting is the rule but the source of

Preliminary Statement to a Theory of Attitudes 463

an item of information is more quickly forgotten than the material pre-
sented" [27, p. 197_. Collier has reported interesting results from
students who were asked to read Xazi propaganda fl8"i. In spite of in-
structions, they came through the task with more favorable attitudes
toward the Nazis than were held by other students. Again, the effect of
contextual factors did not prevent the information absorbed from the
documents from having an influence.

11. Ego-defensive attitudes will be relatively susceptible to change
through procedures designed to give self-insight and will be resistant to
change through procedures employing information and action. Our
major thesis has been that since attitudes serve different needs and func-
tions, they can be changed only through relating the change-procedure
to the appropriate motive pattern. In general this calls for separating
subjects on the basis of their needs and values to begin with and making
differential predictions for various change methods. Thus far the greater
bulk of the research on attitude change has started with the attitude
itself and has assumed a common motive pattern for all people. Our
own method, as shown by the following experiments, is to begin with
measures of ego-defensiveness as one of the major sources of attitudes
toward minority groups and to gear influences directed at change to the
anticipated motive patterns.

Subjects high in ego-defensiveness as measured by projective tests
were not significantly influenced by information and attempts at cog-
nitive restructuring in the experiment of Katz, Sarnoff, and McClintock
[34]. Unfavorable attitudes toward Negroes both with respect to cogni-
tive and behavioral components were not appreciably affected by new
information. These findings were replicated in a follow-up study em-
ploying a different population. Self-insight procedures were also em-
ployed in these studies on the assumption that the people in the middle
ranges of ego-defensiveness would be most responsive to the influence.
High ego defenders, it was assumed, could be affected only by fairly
intensive therapy. McClintock has shown that the insight procedure,
employing a case study illustrative of the dynamics of repression and pro-
jectivity, did in fact produce the anticipated changes [46]. The same type
of defensive people who had resisted the information approach now be-
came more favorable toward Negroes. McClintock further demonstrated
that conformity pressures had little influence upon ego defenders but did
successfully change people who possessed a high degree of the need to
conform.

The success of Culbertson's role-playing experiment in changing atti-
tudes toward Negroes has been mentioned but it is significant that her
important changes occurred among her subjects scoring low in the F
scale [20]. The high ego defenders showed little change as a result of role

464 DANIEL KATZ AND EZRA STOTLAND

playing. Finally, It should be added that Wagman found that people

low in ego-defensiveness resisted authoritarian suggestion directed at
changing their attitudes but responded positively to information [68].
The people high in ego-defensiveness resisted the information but were
influenced by authoritarian suggestion toward more favorable or un-
favorable attitudes toward Negroes.

SUMMARY

The concept of attitude is useful in social psychology if it is not
stretched to cover all aspects of psychological functioning and if it is also
given specifications within the area to which it refers. On the first count,
we have limited attitudes to evaluations of objects and have ruled out
beliefs which are not colored by affect and affective processes which are
not tied to cognitive elements. On the second count, we have described
characteristics of attitudes and their relation to motivational processes and
have suggested a typology of attitudes which takes into account both
structural and functional aspects. This analysis may be helpful in dealing
with problems of attitude formation and change. Not all types of atti-
tudes follow the same pattern of dynamics either in genesis or in subse-
quent modification. Hence a procedure which will be successful in
changing one type of attitude may be completely ineffective in attempt-
ing to change another attitude.

The interactional nature of the factors in social experience makes it
possible for change to be introduced with varying effectiveness into any
part of the psychological system of which the attitude is a part. If there
is a change in the person's needs, in his beliefs, in his values, in his per-
ceptions, or in his behavior, there can be modifications of his attitudes.

At one time or another, social scientists have emphasized a single
factor as the major determinant of attitudes. Some emphasized the social
environment of the individual and made ecology central in their ex-
planation. A related theory made behavior the important variable. Peo-
ple become aware of their roles after they have played them. The mod-
ern emphasis is upon the individual's own perception, upon his "defini-
tion of the situation." We believe that emphasizing one set of factors
produces a model inadequate for dealing with social attitudes. The phe-
nomenological approach does not enable us to predict a fairly common
occurrence in social life, namely, that attitudes develop to justify be-
havior which is imposed upon the individual. Slaves do not generally
hug their chains, but within a wide range of limiting conditions, people
accept their prescribed roles and evaluate them favorably. The tech-
nique of the fait accompli is certainly as old and as common as the
technique of trial balloons.

Preliminary Statement to a Theory of Attitudes 465

In addition to describing characteristics of attitudes, we believe it is
important to include concepts which deal with the relation of attitude
to personality and the relation of attitude to the objective social world.
Hence, we have spoken of the isolation and connectedness of attitudes to
value systems and of the instrumental function of attitudes for the pro-
tection of the self-image and for ego defense. We have also suggested the
notion of the appropriateness of the attitude to the object to which it
refers. Appropriateness resembles the concept of veridical perception.

Though it is possible to enter the psychological system at the point of
need, belief, or behavior, not all attitudes are equally susceptible to in-
fluences directed at beliefs or behavior. Ego-defensive attitudes are par-
ticularly resistant to environmental forces which exert direct pressure to
create change. In this instance, we are dealing with a need within the
individual which is sufficiently complex in origin to make personality
change the necessary condition for modification of the attitude.

The implications of this analysis for research are twofold. The assess-
ment of attitudes should include more than the measurement of affectiv-
ity and evaluation. It should also include measures of the belief com-
ponent, the behavioral component, and the linkage of the attitude to its
value system. Moreover, research on attitudes should assess the motiva-
tional basis of the attitude. Secondly, in research on attitude change, the
procedures to produce change should be designed to affect a specified
factor or factors which previous assessment has suggested as particularly
significant. The change procedure used should represent the manipula-
tion of a known factor of some degree of generality so that general state-
ments are possible about the effect of a given variable upon certain types
of psychological functioning. Finally, research on attitudes should em-
phasize experiments on attitude change, since change is critical for
understanding any phenomenon.

THE RELATION OF THE PRESENT FORMULATION TO THE
ORGANIZATIONAL PLAN OF THIS VOLUME

The foregoing attempt at an initial statement of a theory of attitude
structure and change has not followed the rubrics suggested for theory
presentation. We think that in this area we are not yet far enough along
in the measurement of variables or in systematic conceptualization to
meet the formal demands of many of these categories. To have so stated
our definitions, distinctions, and assumptions might have led to inference
that we are attempting a more fully developed theoretical model than is
the case. On the other hand, it may be helpful to point out the instances
in which the conventional rubrics could be profitably used in the under-
standing and possible further elaboration of our material.

466 DANIEL KATZ AND EZRA STOTLAND

Background Factors and Orienting Attitudes

We would make two observations concerning background factors
which have influenced our thinking. One is the remarkable durability of
the concept of attitudes in social psychology. Most theories concerned
with the social nature of man employ the concept of attitude and often
use the term itself, even though there is little necessary relation between
the formal requirements of the theory and the construct of attitudes. This
suggests that an adequate social psychology must include the concept of
attitude or some very similar construct and that an examination of the
problems in this area is of crucial importance to progress in our field.
We feel that attempts at predicting and understanding behavior in the
social world offer confirmation of this historical conclusion. Efforts to
deal with the real world show our need for a concept more flexible and
more covert than habit, more specifically oriented to social objects than
personality traits, less global than value systems, more directive than
beliefs., and more ideational than motive pattern. Recently there has been
progress along the major fronts of cognition and motivation. We believe
that the study of attitudes is the means most likely to link these two lines
of development.

A second background factor is our impression of the slow progress of
cumulative knowledge in social psychology. Terminology and limited
theories have multiplied, to be sure, and the literature has grown in
mushroom fashion. Yet the advance of knowledge has been discourag-
ingly slow. We believe, therefore, that something may be gained by pur-
suing an old problem and examining it thoroughly before pushing into
areas which seem new but which may be novel only in fresh terms and
labels. The strategy of skimming off the cream and moving on needs to
be supplemented by thorough attention to long-standing problems.

Our major orienting attitudes can be summarized as follows :

1. The prediction of behavior is the major goal of social psychology.
We see very limited value in highly general theories which can explain
after the fact but contain no specifications for the prediction of social
events. We do not mean that the scientist should attempt to predict the
phenotypical event in its uniqueness. But reaction against the phenotypic
can become an excuse to avoid critical predictions.

2. Psychological prediction must concern itself more with major
variance in social behavior. We must walk before we can run, to be sure,
but there are disadvantages in the complete absorption in any difference
which is statistically significant even though it may account for a very
small fraction of the total variance. One consequence is a failure to Con-
firm findings when studies are replicated. Another is the development

Preliminary Statement to a Theory of Attitudes 467

of a social psychology Hmlted to certain restricted aspects of classroom
situations.

3. Models should be helpful in the development of any science, but
formal mathematical models have thus far made very limited contribu-
tions to the progress of social psychology. One reason may be that model
building becomes an end in itself. The tendency is to forget what is al-
ready known and to start with such oversimplified schemes that they
generate no meaningful predictions for social events. Another reason may
He in a misconception of the role of models. They can help in the sys-
tematization and logical ordering of knowledge and in derivations which
might otherwise be neglected. But they do not in themselves give us
major theoretical insights or lead to significant discoveries. Models should
serve a substantive theory rather than replace theorizing about the na-
ture of psychological processes. The models frequently used are borrowed
from other fields and so do not take the place of concepts derived from
a direct study of the phenomena of our own field. Part of the vitality of
Freudian concepts stems from the fact that they grew out of observa-
tion and speculation about human behavior and wrere not attempts to
borrow7 from physics or physiology.

The limited type of model now popular in psychological theorizing
is adapted to laboratory experimentation where a problem is narrowed
down to a few variables which permit fairly precise measurement. But
social psychology requires a model more appropriate to field studies.
Most of the other sciences have grown because of their ability to produce
within the laboratory powerful manipulations which are called for by
their theories. Social psychology is handicapped in this respect because,
for the most part, it can create only relatively weak variables within the
confines of the laboratory. The laboratory approach should be utilized
as fully as possible, but we will make merely limited progress until we can
study the powerful forces which affect people in the real social world.
The systematic study of social change calls for models appropriate to
this level of investigation and such appropriate models are not now avail-
able. F. H. AHport's event system theory of behavior is a promising
development in this direction.

4. Besides rigorous hypothetico-deductive axiomatization, currently,
social psychology needs some intermediate level of systematic concepts
which have particular relevance for the prediction of social behavior.
We need hunches about the significant variables and combination of
variables which produce movement or change in the social world. Weak
as were the formulations of Emile Durkheim, Max Weber, Karl Marx,
and Sigmund Freud from the standpoint of formal scientific theory, we
still lack a systematic set of concepts as content-oriented and as useful as

468 DANIEL KATZ AND EZRA STOTLAND

those provided by these early writers. Small wonder that when we have
to deal with social reality, we revive concepts of anomie from Durkheim,
legitimacy, charisma, and bureaucratic structure from Weber, and pro-
duction relations, powTer, and group conflict from Marx.

This is not to take the position of the man in the street, who wants
science to give him immediate answers to all practical problems. We do
hold, however, that a major referent of our science is the ongoing stream
of social events and that our major need is some system of constructs
which would enable us to move in and measure these ongoing forces.
Social psychology would be better off with a fairly loose system of con-
cepts, such as those provided by Freud for a study of personality, than
with a rigorous formal system which will not permit coming to grips with
social realities nor give any basis for predicting social occurrences.

Structure of the system: independent, intervening, and dependent
variables. The variables we have discussed form an interdependent sys-
tem which can be entered at any point. The causal sequence can flow
in either direction between two variables and lead to circular reinforce-
ment. A form of behavior imposed by environmental forces can lead to
a set of beliefs and these beliefs can, in turn, result in the behavior in
question. We do assume, however, some priority of variables in the deter-
mination of behavior. Though beliefs can modify needs and can affect
their means of satisfaction, we would in general regard motives and en-
vironmental forces as independent variables, attitudes as intervening
variables, and their expression in behavior as the dependent variable. We
shall attempt to apply this ordering to the following motive patterns basic
to attitude formation :

Ego-defensive attitudes toward out-groups. In line with the theo-
rizing of Frenkel-Brunswik and her associates, we regard the defense
mechanisms of repression and projectivity, in combination, as the inde-
pendent variables leading to the intervening variable of an ego-defensive
attitude, with the dependent variable being prejudiced behavior which
could assume the form of ( 1 ) discriminating and aggressive acts toward
out-groups, (2) negative stereotypes of out-groups, and (3) expressions
of negative affect toward out-groups. All three forms of behavior would
be predicted for this type of ego-defensive attitude. The independent
variable could be measured by TAT protocols or the Michigan Sentence
Completion Test, the MMPI, the Blacky test, and certain portions of the
F scale. The dependent variable could be measured by observation of
behavior in natural settings or by stereotype and attitude scores on
questionnaires or in interviews. But not all prejudiced behavior would
derive from ego-defensive attitudes. Moreover, the ego-defensive attitude
would always be accompanied by fairly intense feelings of hostility either
in the expression of negative stereotypes or of discriminating acts. In

Preliminary Statement to a Theory of Attitudes 469

other words, the behavior could properly be considered aggressive. Dis-
criminator}* behavior and the expression of negative stereotypes which
do not involve high affect would have other origins and functions than
ego-defensiveness. They could be object-instrumental or ego-instrumental
(see discussion below).

The ego-defensive attitude of conformity. Another pattern of inde-
pendent variables would combine the defense mechanism of conformity
and the existence of clear social norms of the groups in which the indi-
vidual moves. Again, a type of ego-defensive attitude is postulated as the
Intervening variable, and the dependent variable becomes the expression
of this attitude in the approval of the practices and beliefs sanctioned by
group norms. In this case the pattern of independent variables would be
measured as follows: the defense mechanism of conformity would be
assumed if there is weak ego strength plus a high score on other-directed-
ness. The Morris Paths of Life is a possible measure of other-dlrectedness
and the Thomas-Zander-Stotland scale is a possible measure of ego-
strength. Not all conformity, even as a generalized trait, is ego defensive.
This is the reason for including a measure of ego strength with a meas-
ure of other-dlrectedness. The social norms would have to be ascertained
through a statistical survey of a representative sample of the relevant
group population.

Proximal attitude. The independent variable here would be an ob-
ject in the environment which gives the individual consummatory satis-
faction for some one of his motives. The other variable in this pattern
would be a need state which arouses the motive in question. The Inter-
vening variable would be the proximal type of attitude, and the depend-
ent variable would be ( 1 ) the expression of favorable beliefs about the
object, (2) the expression of favorable affect toward it, and (3) patterns
of overt behavior to acquire the object. Favorable beliefs and favorable
affect could be expressed without the arousal of the motive but acquisi-
tive behavior would require motive arousal. The measure of the inde-
pendent variable would depend upon the motive in question. For a
relatively uncomplicated appetitive drive like hunger, the individual
could be asked such questions as "How hungry are you?33 "When did
you eat last?33 "What foods do you like?33 Other more complex motives
would have to be measured either through longitudinal observation of
the person or through projective tests similar to the measures for need
achievement, need affiliation, and need for power.

Object-instrumental attitudes. The independent variables here would
be objects which are instrumental to motive satisfaction but nonconsum-
matory and be the arousal of the relevant motive. The intervening variable
would be an object-instrumental attitude; the dependent variable would
be the expression of favorable affect toward the object, the expression of

470 DANIEL KATZ AND EZRA STOTLAND

favorable beliefs toward it, and other positive behavior toward it. Again,
the affect and the expressions of belief could be elicited without motive
arousal, but the full pattern of overt behavior of an instrumental sort
would depend on motive arousal. The measures here would follow the
same pattern as suggested for proximal attitudes.

Ego-instrumental attitudes. The independent variable here wrould be
an object which is instrumental to the satisfaction of ego motives such
as ego enhancement and self-determination. The intervening variable
would be the ego-instrumental attitude. The dependent variable would
consist of a favorable evaluation of the object. The measure of the inde-
pendent variable would be based upon tests of objects which enhance the
ego as indicated either by the self-report of the subject or by the reports
of outside observers.

Affective associations. The independent variable would be an object
which has been involved in the satisfactions of a motive but which bears
no necessary relationship to such satisfaction. The intervening variable
would be affective associations, and the dependent variable would be
affect expressed toward the object, behavioral avoidance of the object
if the association had been unpleasant but no positive overt pattern of
behavior if the association had been pleasant. The measure of the inde-
pendent variable could be based upon longitudinal observation of the
individual or could be derived from statistical norms of the relevance of
objects for motive satisfaction for many subjects who came from the
same background.

The independent variables listed above all relate to the processes of
need satisfaction in the person. Another type of dynamic sterns from the
tendency of the components of the attitude (affective, cognitive, and be-
havioral) to be consistent with each other. Thus, a change in one com-
ponent can lead to a change in another. The first component is, there-
fore, an independent variable, whereas the other is dependent. We
assume that the affective component has the most potency in changing
other components and the cognitive has the least. The affective com-
ponent can, of course, be influenced by changes in the person's pattern
of need satisfaction, as indicated above. Such changes in need satisfac-
tion act as independent variables to produce changes in all three com-
ponents through first influencing the affective component. A person may
sometimes behave toward the object of his attitude in quite a different
fashion than would be predicted from a knowledge of the attitude. If
this behavior is the result of some sustained environmental force from
which the individual cannot escape, the behavioral component of the old
attitude will become more consistent with the expressed behavior. As a
result, the other components of the attitude will change to become more
consistent with the modified behavioral component. The repeated oaths

Preliminary Statement to a Theory of Attitudes 471

of loyalty to Hitler by German officers helped to modify attitudes toward
the Fiihrer. In this type of attitude change, the independent variable is
the behavior toward the object and the components of the attitude are
the dependent variable.

Finally, we would admit, but give lowest priority, to changes which
come about through changes in the cognitive component. A person may
be given new information which changes his beliefs about an object, and
in turn his feelings about it change as does his action orientation.

Barriers Blocking General Theoretical Advance in Psychology

We have already indicated our belief concerning the need for an
intermediate level of concepts which would have some content orienta-
tion. We hold that one of the real barriers to general theoretical advance
in social psychology is the distance between genotypic constructs and our
phenotypic measures. In physics, the concept of atmospheric pressure is
fairly close to its operational measurement. In physiological psychology,
many concepts are similarly tied to their operational measurement. In
personality theory and in social psychology, however, concepts like ego
strength, defense mechanisms, role systems, and role conflict are so re-
mote from their measurement that we have no single, clearly required
set of operational measures.

We believe this is a basic difference between the social and the nat-
ural sciences. In general, this separation of concepts and their phenotypic
indicators has produced two consequences. Factually minded investi-
gators have pursued phenotypic observations and measures and have
given us rank empiricism. In the natural sciences, such rank empiricism
would have been much more useful because generalization would have
emerged readily from the collection of facts. Theoretically minded in-
vestigators, on the other hand, have been satisfied with any measure
which could be remotely justified as an indicator of the concept with
which they were concerned. Thus their research has not led to cumu-
lative knowledge. It is not because behavioral scientists are essentially
different from natural scientists that there has been less progress in the
behavioral sciences. It is because the closer relationship between the con-
cepts and phenotypic measures in the natural sciences imposes objective
restraints upon the investigator. We will not make substantial progress in
the behavioral sciences unless we recognize the barrier produced by the
nature of our subject matter and attack it along two major fronts: the
development of an intermediate level of concepts between our genotypic
constructs and their phenotypic indicators, that is, concepts of some level
of generality which still point toward a class of objects; and more sys-
tematic testing of all the assumptions of a theoretical scheme, including
the exploration of its relationship to a variety of empirical settings.

472 DANIEL KATZ AND EZRA STOTLAND

REFERENCES

1. Adorno, T. W., Frenkel-Bmnswik, Else, Levinson, D. J., & Sanford,
R. N. The authoritarian personality. New York: Harper, 1950.

2. Allport, F. H. Social psychology. Boston: Houghton Mifflin, 1924.

3. Allport, F. H., & Hartman. D. A. The measurement and motivation
of atypical opinion in a certain group. Amer. pol. scL Rev.y 1925, 19, 735-
760.

4. Allport, G. W. Personality. London: Constable, 1937.

5. Allport G. W. Attitudes. In C. M. Murchison (Ed.), Handbook of
social psychology. Worcester, Mass.: Clark Univer. Press, 1935. Pp. 798-
844.

6. Asch, S. E. Social psychology. Englewood Cliffs, N.J.: Prentice-Hall,
1952.

7. Ashley, W. R., Harper, R. S., & Runyon, D. L. The perceived size of
coins in normal and hypnotically induced economic states. Amer. J. Psychol.,
1951, 64, 564-572.

8. Back, K. W. Influence through social communication. /. abnorm. soc.
Psychol, 1951, 46, 9-23.

9. Bain, R. An attitude on attitude research. Amer. ]. Sociol, 1928, 33,
440-457.

10. Bettelheim, B.3 & Janowitz, M. Dynamics of prejudice. New York:
Harper, 1950.

11. Brownfain, J. J. Stability of the self-concept as a dimension of
personality. /. abnorm. soc. Psychol, 1952, 47, 597-606.

12. Carlson, E. R. Attitude change through modification of attitude
structure. /. abnorm. soc. Psychol, 1956, 52, 256-261.

13. Cartwright, D. Some principles of mass persuasion. Hum. Relat.,
1949, 2, 253-267.

14. Cartwright, D., & Harary, F. Structural balance: a generalization
of Heider's theory. Psychol Rev., 1956, 63, 277-293.

15. Charters, W. W., Jr., & Newcomb, T. M. Some attitudinal effects of
experimentally increased salience of a membership group. In G. E. Swanson,
T. M. Newcomb, & E. L. Hartley (Eds.), Readings in social psychology.
(Rev. ed.) New York: Holt, 1952. Pp. 415-419.

16. Clark, K. Desegregation, an appraisal of the evidence. /. soc. Issues,
1953,9 (4).

17. Cohen, A. R., Stotland, E., & Wolfe, D. M. An experimental in-
vestigation of need for cognition. /. abnorm. soc. Psychol. 9 1955, 51, 291-
294.

18. Collier, R. M. The effect of propaganda upon attitude following a
critical examination of the propaganda itself. /. soc. Psychol. 3 1944, 20,
3-17.

19. Crockett, W. H. The effect on attitude change of cognitive dif-
ferentiation and affect, under conditions of norm presentation with and
without counter-arguments. Unpublished doctoral dissertation, Univer. of
Michigan, 1952.

Preliminary Statement to a Theory of Attitudes 473

20. Culbertson. Frances Mitchell. The modification of an emotionally-
held attitude through role playing /. abnorm. soc. Psychol., 1957. 545 230-
233.

21. Edward, A. L. Political frames of reference as a factor in influencing
recognition. /. abnorm. soc. Psychol., 1941, 36, 34-61,

22. Festinerer, L. A theory of social comparison process. Hum. Relat.,
1954, 83 117-140.

23. Harding, J., & Hogrefe, R. Attitudes of white department store
employees toward Negro co-workers. /. soc. Issues, 1952, 8 fl}, 18-28.

24. Hartmann, G. W. A field experiment on the comparative effective-
ness of "emotional" and "rational" political leaflets in determining election
results. /. abnorm. soc. Psychol., 1936, 31, 99-114.

25. Hartley, E. L. Problems in prejudice. New York: King's Crown
Press, 1946.

26. Heider, F. Attitudes and cognitive organization. /. Psychol, 1946,
21, 107-112.

27. Hovland, C. I., Lumsdaine, A. A., & Sheffield. F. D. Experiments on
mass communication, Princeton, N.J.: Princeton Univer. Press, 1949.

28. Hovland, C. I., & Weiss, W. The influence of source credibility on
communication effectiveness. PubL Opin. Quart., 1951, 15, 635-650.

29. Janis, I. L., & Feshbach, S. Effects of fear-arousing communications.
/. abnorm. soc. Psychol., 1953, 48, 78-92.

30. Janis, I. L., & King, B. T. The influence of role-playing on opinion
change. /. abnorm. soc. Psychol., 1954, 49, 211-218.

31. Janis, I. L., Lumsdaine, A. A., £ Gladstone, A. I. Effects of pre-
paratory communication on reactions to a subsequent news event. PubL
Opin. Quart., 1951, 15, 487-518.

32. Katz, D., & Allport, F. H. Students' attitudes. Syracuse, N.Y.:
Craftsman, 1931.

33. Katz, D., & Kahn, R. L. Some recent findings in human-relations
research in industry. In G. E. Swanson, T. M. Newcomb, & E. L. Hartley
(Eds.), Readings in social psychology. New York: Holt, 1952. Pp. 650-665.

34. Katz, D., Sarnoff, I., & McClintock, C. Ego defense and attitude
change. Hum. Relat., 1956, 9, 27-46.

35. Kelman, H. G. Three processes of acceptance of social influence:
compliance, identification and internalization (mimeo). National Institute
of Mental Health, 1956.

36. Kelman, H. C., & Hovland, C. I. "Reinstatement" of the com-
munication in delayed measurement of opinion change. /. abnorm. soc.
Psychol., 1953, 48, 327-335.

37. Knower, F. H. Experimental studies of changes in attitude. I. A
study of the effect of oral argument on changes of attitudes. /. soc. Psychol.,
1935, 6, 315-347.

38. Knower, F. H. Experimental studies of changes in attitude. II. A
study of the effect of printed argument on changes in attitude. /. abnorm,
soc. Psychol., 1936, 30, 522-532.

474 DANIEL KATZ AND EZRA STOTLAND

39. Kramer, B. M. Dimensions of prejudice. /. Psycho!., 1949, 27, 389-
451.

40. Krech, D., & Cratchfield, R. S. Theory and problems of social psy-
chology. New York: McGraw-Hill, 1948.

41. Kutner, B.3 Wilkins, Carol, & Yarrow, Penny R. Verbal attitudes
and overt behavior involving racial prejudice. /. abnorm. soc. PsychoL,
1952, 47, 649-652.

42. LaPiere, R. T. Attitudes vs. actions. Soc. forces, 1934, 13, 230-237.

43. Lazarus, R. S., & McCleary, R. A. Autonomic discrimination with-
out awareness: a study of subception. PsychoL Rev., 1951, 58, 113-122.

44. Levine, J. M.? & Murphy, G. The learning and forgetting of contro-
versial material. /. abnorm. soc. PsychoL, 1943, 38, 507-517.

45. Lieberman, S. The relationship between attitudes and roles: a
natural field experiment. Unpublished doctoral dissertation, Univer. of
Michigan, 1954.

46. McCiintock, C. G. Personality syndromes and attitude change. /.
Pers., 1958, 26, 479-493.

47. McDougall, W. Introduction to social psychology. Boston: Luce,
1908.

48. Miller, N. Learnable drives and behavior. In S. S. Stevens (Ed.),
Handbook of experimental psychology. New York: Wiley, 1951. Pp. 435-
472.

49. Murray, H. A. The effect of fear upon estimates of the maliciousness
of other personalities. /. soc. PsychoL, 1933, 4, 310-329.

50. Newcomb, T. M. Social psychology. New York: Dryden, 1950.

51. Newcomb, T. M. An approach to the study of communicative acts.
PsychoL Rev., 1953, 60, 393-404.

52. Osgood, C. E., & Suci, G. J. Factor analysis of meaning. /. exp.
PsychoL, 1955, 50, 325-338.

53. Osgood, G. E., & Tannenbaum, P. H. The principle of congruity in
the prediction of attitude change. PsychoL Rev., 1955, 62, 42-55.

54. Peak, Helen. Attitude and motivation. In M. R. Jones (Ed.),
Nebraska symposium on motivation 1955. Lincoln, Neb.: Univer. Nebraska
Press, 1955. Pp. 149-188.

55. Raven, B. H. The effect of group pressure on opinion, perception,
and communication. Unpublished doctoral dissertation, Univ. of Mich-
igan, 1953.

56. Razran, G. H. S. Conditioning away social bias by the luncheon
technique. PsychoL Bull., 1938, 35, 693.

57. Riesman, D. The lonely crowd. New Haven, Conn.: Yale Univer.
Press, 1950.

58. Rosenberg, M. J. Cognitive structure and attitudinal affect. /. ab-
norm. soc. PsychoL, 1956, 53, 367-372.

59. Schanck, R. L. A study of a community and its groups and in-
stitutions conceived of as behavior of individuals. PsychoL Monogr., 1932,
43 (2).

Preliminary Statement to a Theory of Attitudes 475

60. Seashore, S. E. Group coheszveness in the industrial work group.
Ann Arbor, Mich.: Inst. for Social Research, 1954.

61. Shand, A. F. The foundation of character. London: Macmillan,
1926.

62. Smith, M. B.3 Bmner, J. S., & White, R. W. Opinions and per-
sonality. New York: Wiley, 1956.

63. Spranger, E. Types of men. Halle: Niemy, 1928.

64. Stouffer, S. A. An analysis of conflicting social norms. Amer. social
Rev., 1949, 14,707-717.

65. Suchman, E. A. The intensity component in attitude and opinion
research. In S. A. Stouffer, et a!., Measurement and prediction. Princeton,
N.J.: Princeton Univer. Press. 1950. Pp. 213-277.

66. Symonds, P. M. What is an attitude? Psychol. Bull, 1927, 24, 200-
201.

67. Thurstone, L. L., & Ghave, E. J. The measurement of attitude.
Chicago: Univer. Chicago Press, 1929.

68. Wagman, M. Attitude change and the authoritarian personality. /.
PsychoL, 1955, 40, 3-24.

69. Watson, J. B. Behaviorism. New York: Norton, 1925.

70. Watson, W. S., & Hartmann, G. W. The rigidity of a basic attitudinal
frame. /. abnorm. soc. Psychol, 1939, 34, 314-335.

71. Wertheimer, M. Productive thinking. New York: Harper, 1945.

72. Woodworth, R. L. Dynamic psychology. New York: Columbia
Univer. Press, 1918.

LATENT STRUCTURE ANALYSIS

PAUL F. LAZARSFELD
Columbia University

Introduction 477

I. Some Problems of Classification in the Social Sciences {1} 479

The trait and other intervening variables 479

The disposition concept 484

The property space 487

The relation of manifest to latent property space 489

II. The Logical Foundation of Latent Structure Analysis: A Synopsis of the

Main Issues {2, 3, 6} 491

Some initial clarifications 491

Item analysis and item curves 494

Item analysis applied simultaneously to more than one item . . . .495
The "mixture" phenomenon and its role in the explanation of statistical

relations 498

The accounting equations and the principle of local independence . . 502

Accounting equations 503

III. The Nine Steps of Latent Structure Analysis {4, 5} 506

Summary of the nine steps 506

Step 1 : Choice and specification of the model 508

Step 2: Accounting equations specialized for the model 509

Step 3 : The conditions of reducibility 512

Step 4; Identifiability 515

Step 5: Identification 516

Step 6: Computation: The fitting procedure 518

Step 7: Evaluation of the fit 523

Step 8: The recruitment pattern 525

Step 9: Classification and scores 527

Summary 527

IV. The Promises and Limitations of Latent Structure Analysis {7—12} . . 528

The meaning of trace lines 529

Distribution in the latent space 532

476

Latent Structure Analysis 477

Comparison with factor analysis 538

Comparison with test theory 540

References 542

INTRODUCTION

All the social sciences deal with concepts which seem somewhat
vague. Who can, in practice, recognize an extrovert personality? Who
has not read many discussions as to the real meaning of public opin-
ion? \Vho can say precisely what a folk society is? There are various
reasons why the social scientists' language has so many of these terms,
which at first sight seem to be ill defined and even at their best are
"fuzzy at the fringe.35 In some cases wre can, by the nature of the con-
cept, only observe symptoms, behind which we assume a more perma-
nent reality. This would be true, for example, in the case of personality
notions. In other matters the object of investigation is so vast that we
can analyze only certain aspects of it: notions like patterns of culture
or Zeitgeist belong here. For still other purposes the problem itself
seems to require a looser kind of formulation: wherever we study
adjustments — e.g., in marriage, in job performance, or in standard
of living — we find that large numbers of actual solutions may serve
the same functional purpose.

This peculiarity of the social scientist's intellectual tools has been
deplored by some, considered as unavoidable by others. Most of all,
however, it has been covered with nomenclature. Syndromes, geno-
types, underlying concepts, hypothetical constructs, and many other
terms have been used. It is hard to say to what extent we have today
a clear formulation of the problem behind all these terms, let alone
clear directions on how to deal with them in the pursuit of empirical
research. And yet it is in the course of actual investigations that some
clarification is most needed. For if we have to decide whether there is
increased bureaucratization in government, or whether city life makes
people progressively neurotic, we must get some measures of these
tendencies. And whatever index we use, we make implicit assumptions
about the meaning of the kind of terms which we have just exemplified.

Thus, problems of measurement, of meaning, and of concept
formation fuse necessarily into each other. No empirical procedure
of classifying social objects can be understood without reference to
general logical discussions. And philosophical disquisitions about
the nature of the social sciences are not likely to be fruitful without an
incisive analysis as to how empirical social research does actually
proceed.

The present paper will analyze one special procedure by which
it is possible to make what one might call inferential classifications.

478 PAUL F. LAZARSFELD

Any number of well-known topics are covered by this provisional
name tag: a person's attitude as inferred from his behavior, the inten-
tion of a document as inferred from certain linguistic characteristics,
the morale of a group as inferred from its various performances, and
many others. No exhaustive listing or explicit definition will be given
of the applications which we intend to cover. If it were possible to
state clearly at the beginning the purpose of the procedures to be
described, the whole paper — even the procedure itself — might be
superfluous. But the basic thesis to be developed is exactly this:
measurement, classification, and concept formation in the social
sciences exhibit special difficulties; they can be met by a variety of
procedures, and only a careful analysis of the procedure and its rela-
tion to alternative solutions can clarify the problem which the pro-
cedure attempts to solve.

The editor of this series of monographs developed a careful discus-
sion of the nature of a "systematic formulation" for the contributors.
He had in mind theoretical developments which attempted prediction
of substantive observation. As will be seen, the following pages deal
with the organization of such data. Consequently his outline could not
be followed in detail. The presentation which our topic required turns
out in retrospect, however, not to fit badly the general program of the
whole project.1

The first section, indeed, deals with "background factors and orient-
ing attitudes'3 {!}. It takes some problems well known to psychologists
and briefly sketches how they became clearer and more articulate
between, say, 1900 and 1930. It shows how logicians have looked at the
same matter. Finally, it suggests a formulation which leads to the
threshold of the solution offered by latent structure analysis.

The second section turns to the "structure of the system35 O and to
the "initial grounds for its assumptions3 * -OK It shows that a central
axiom of local independence is an idealization of certain empirical
procedures habitual in test construction and in survey analysis. The
distinction between manifest data and latent parameters takes the
place of the three types of variables suggested by the general outline.
The accounting equations presented toward the end of this section
summarize the "formal organization of the system" -C6K

The third section carries one example through the nine basic
steps of a latent structure analysis. It gives, thus, an idea of the "meas-
urational and quantificational procedures" *5>. At this point the limita-
tion of space is most obvious. The main contribution of the whole
approach is its mathematical flexibility. The "construction of func-

1 In the following paragraphs we refer to the rubrics of the editor by citations fol-
lowed by the appropriate number.

Latent Structure Analysis 479

tion forms55 i4> depends upon the data under Investigation, and their
adequacy can be tested empirically. But each type of accounting
equation must be studied In its own right. One example, therefore,
can illustrate only the main principles Involved and cannot provide
procedures applicable to all models.

The fourth section does not continue the mathematical discussion
but gives. Instead, a variety of findings to Indicate the "range of ap-
plications55 C7K The "mediating function53 i8> of latent structure
analysis is approached by discussing Its relation to factor analysis and
to formal test theory. It Is not easy to decide whether one can talk of
"evidence for the system53 i9> In what is essentially a theory of concept
formation. But even In terms of usefulness, it Is too early to form a judg-
ment. At the same time, It would be easy to show that "outside the
present context55 i!0>, that is, beyond traditional psychology, the pro-
cedure has numerous applications to sociological and anthropological
problems. No attempt was made to digress along these lines. The sec-
tion, however, gives an idea of the high "degree of programmatlclty53
ill) the system has and how the "intermediate and long-range strategy53
{12> points to numerous specific problems, which are not yet solved and
require the collaboration of mathematicians, experimenters, and
behavioral theorists.

A word needs to be added as to the administrative history of latent
structure analysis. Much of the early work was generously supported
by the Rand Corporation. An over-all monograph has been in draft
for several years but was always delayed by new developments. In the
meantime, some of the publications listed among the references have
appeared. Other findings listed there are available as dissertations at
Columbia University. A number of results, however, are still in-
corporated only in memoranda to the Rand Corporation. To facilitate
orientation, the text and the references mention only two such
memoranda. The first is a summary of the more recent mathematical
developments; the second a collection of specific studies, in which
various models were applied to empirical data. Finally, the author
wishes to thank Mr. Arnold Simmel for much editorial and computa-
tional help and Miss June Alter for resourceful secretarial work in the
preparation of the present summary.

I. SOME PROBLEMS OF CLASSIFICATION IN THE SOCIAL SCIENCES

The trait and other intervening variables. Traits became the
topic of more systematic reflection in connection with moral problems
— at least as far as American psychological literature goes. One starts
naturally with William James5s Principles of Psychology; there one does

480 PAUL F. LAZARSFELD

not find the term cc trait" In the index. In the chapter on habits,
however (chap. 10), there is an extensive discussion of how people can
acquire desirable habits like industriousness, or lose undesirable ones
like drunkenness. This Intertwining of ethical problems with the ques-
tion of traits, their acquisition and change. Is still equally strong In
John Dewrey; it is the main theme of his Human Nature and Conduct.
Dewey also uses "habit" as his central term; he considers it Inter-
changeable., however, with terms like trait, characteristic, attitude,
and tendency. Dewey5 s concern with the changing of habits, the exam-
ples he chooses, and the advice he gives are often very similar to
James's treatment.

The moralist observes differences in his and other people's conduct,
tags them as good or bad, and reflects on how valuable traits can be
strengthened. The methodologist starts from the same observations but
is more Interested in defining, classifying, and measuring these traits.
The pragmatists were a combination of moralist and logician, and
they found a way to fuse their double motivation into a view which
combined their activistic philosophy and their operational idea of
scientific work. James showed this clearly [11, our italics].

Suppose, e.g., that we say a man is "prudent." Concretely, that means that he
takes out insurance^ hedges in betting, looks before he leaps. . . . As a constant habit
in him, a permanent tone of character, it is convenient to call him prudent in
abstraction from any one of his acts. . . . There are peculiarities in his
psychophysical system that make him act prudently. . . .

We are not surprised to see that James is very explicit on the rela-
tion between an inferential concept and the indicators connected with
it. He is, after all, the one who coined the phrase "concepts signify
consequences." He felt that abstract descriptions are often useful
enough, yet they are "sucked up and absorbed without residuum into
the concrete ones, and contain nothing of any essentially other or
higher nature which the concrete descriptions can be justly accused
of leaving behind.35 Here is picturesque language, the precise meaning
of which is not easily checked. But it is safe to assume that James in-
clined toward identifying the concept and its indicators. Now this leads
to obvious difficulties. Do all prudent people always look before they
leap? Where do we take account of the amount of insurance taken
out? How about people who show some but not all the symptoms
mentioned by James? Inversely, should not other indicators have been
included?

Dewey was obviously aware of these difficulties. As a moral philos-
opher he was not less convinced than James of the mutual interaction
between "disposition" and "doing"; today's tradition of progressive

Latent Structure Analysis 481

education Is testimony to his point of view. But as a logician, he saw a
looser connection between concepts and indicators. He warned that
one should not "assume thai there is or ever can be an exact equation
of disposition and outcome.53

It was characteristic of habits [6] "... that their outworking In
any particular case is subject to contingencies, to circumstances which
are unforseeable and which carry an act one side of its usual effect.35

Here a new idea comes to the fore, although rather shadowy at first
— the notion of probability. It is easy to see how the progress from
James to Dewey can be reformulated. The prudent man is likely to look
before he leaps because any specific behavior item Is only a probable
but not a necessary condition of a related trait. This has remained ever
since as one well-recognized element in all presentations. In an early
systematic discussion Allport tried "with the aid of eight criteria to
define trait and to state the logic and some of the evidence for the ad-
mission of this concept to good standing in psychology.33 One of his
criteria is important here. [2]

Acts and even habits that are inconsistent with a trait are not proof of the
non-existence of the trait . . . there are in every personality instances of acts
that are unrelated to traits, the product of the stimulus and of the attitude of the moment.
Even the characteristically neat person may become careless in his haste to
catch a train.

A second element has not found equally clear recognition. How
does the notion of a trait develop, either in daily life or in scientific
work? We experience, say, anxiety, and its role in our own course
of action (R). We observe how other people act in situations (S)
which would, we know, bring on our anxieties; we notice that their
reaction R is similar to ours. As a result, we file away in our minds that
as a rule such stimuli S are likely to be followed by responses R. We
"explain" such S-R sequences with the help of an intervening varia-
ble: anxiety. The value of this construct becomes particularly apparent
if many S-R situations are observed where the S and the R vary, but
where the same intervening variable (anxiety) seems appropriate. We
can then organize our observations in a somewhat more economical
way: we remember the series of x situations which create anxiety and the
series of y responses by which anxiety is expressed. Instead of register-
ing x times y relationships of the S-R type, we need only remember
x + y findings: the x prompters to and they indicators of anxiety.
A schematic presentation, on page 482, has been proposed [18],
If we want to "create anxiety33 we would choose one or more situa-
tions on the left side of the frame, and if we want to "measure anxiety5'
a selection and combination of responses from the right side would be

482 PAUL F. LAZARSFELD

necessary; they would be used as indicators. The important point is
that usually several indicators would be derived by an investigator,
Le,3 facial expressions, interpretations of ink blots, etc. But because
indicators have only a probability relation to traits, the crucial prob-
lem arises as to how they can be combined if they do not all go in the
same direction, for example, if in a specific case RI makes high anxiety
and R*>, low anxiety probable. It is too early even to hint at an answer
here; but it is surprising how grievously the issue can be missed by

Situations z / \ Responses

FIG. 1.

scholars when they leave the field of their special research experience.
When Edward Tolman developed the notion of intervening vari-
ables, he was concerned about how the intervening variables are
operationally defined, measured, or whatever other terms he used to
apply in pointing to the problem. The answer Tolman gave fifteen
years ago [29], when he was concerned with the behavior of a rat at a
choice point, is the same he gave recently when his whole system was
"elaborated for the special case of a hungry actor going to a particular
restaurant and ordering and eating a particular food.5' For the rat it
read as follows [28, p. 333]:

By an operational definition of an intervening variable I shall mean, first,
a statement about a standard defining experiment in which a certain measur-
able variation in some feature of the observed behavior will, by definition, be
assumed to be a direct measure of corresponding variation in the magnitudes of
a given intervening variable. Second, such a definition will involve an assump-
tion about the linear or nonlinear nature of this mathematical function
connecting the measured feature of the dependent behavior to the intervening
variable. And, third, the specific constants in this form of mathematical
function must also be known, or assumed, before such definitions will be final.

The idea is that we can find one specific indicator for each inter-
vening variable. Everything else being constant, the variations in the
indicators correspond to the variations in the intervening variable.

Latent Structure Analysis 483

We have grave doubts whether such a procedure Is feasible even with
animal experiments. And we are confident that It Is the wrong Idea
as far as the study of human behavior Is concerned. Tolman's own
description of how he would proceed In a concrete research situation
shows this. At one point he exemplifies an "actor's belief-value matrix"
by the opinion he has about various restaurants, his food preferences,
etc. How will Tolman find out about this intervening variable? By
[28, p. 295]

. . . mere questionnaires or interviews. Thus, for example, one could ask the
subjects: (1) "What are you ready to do when you haven't eaten for a con-
siderable length of time?" (2) "What kinds of foods do you like? Name six
varieties of food In order of preference. What do you like about each of these
six?55 (3) "For each of these six foods what types of restaurant would you go
to and in what order? List aU the considerations you would take Into account
in choosing the one kind of restaurant or the other.'5

The repeated references to questionnaires In this monograph make
it easy to predict what problems Tolman would face if he were really
to develop measurements along his line of argument: the student ex-
perienced In social research knows that answers to questionnaires
vary considerably if wordings are slightly changed, if the interview
is done under slightly varying conditions, etc. There is just no way to
develop a "standard experimental setup" or "standard defining
experiment.53 We will have to face the fact that to an intervening variable
there will correspond a variety of indicators and that they will have to be reconciled
in some way.

The discussion on intervening variables covers a wide range, and
we cannot enter it in detail. Tolman was chosen as an example
because he is, to our knowledge, the only one who applied the equip-
ment of the learning theorist to an everyday life situation. There exists,
however, a careful analysis of Hull's writing which brings out one
point of special importance in the present context. Koch has pointed
out that the relation between an intended measure and the multiple
ways it can actually be approached does not pertain only to interven-
ing variables in the narrow sense [13]. In discussing the notion of
"independent variables/5 he makes a distinction between systematic
and experimental independence. The "experimental independent
variables may be specific singular realizations of a systematic inde-
pendent variable; they are not however to be identified with it," and
well might not be singular [13; p. 28]. Koch, too, is critical of the
"remote leaps from the data of single defining experiments to general
theoretical statements." He states as a general principle that "all
alternate experimental variables to which a given independent vari-

484 PAUL F. LAZARSFELD

able is reducible . . . must be brought to converge by appropriate
scaling techniques" [13; p. 65]. In a way, Koch's position is very similar
to the one we take here: every "systematic variable" is intervening in
the sense that additional assumptions are needed to link it with an
array of actual observations.

We find then that a tradition has growrn in psychology whereby the
intended classification required by concepts like trait, attitudes inter-
vening variable, etc., is performed by using indicators directly acces-
sible to the investigator. These indicators are presumed to have a
probability relation to the "underlying" (intended) variable; and
because of it, if we use — as we invariably do — a number of indicators
simultaneously, we will always get into "contradictions53 which have
to be reconciled. Before discussing the matter further we shall show
that the psychologists3 problem is only a special case of the general
logical issue regarding disposition concepts.

The disposition concept. In recent writings of logicians, one can
find frequent discussions of "disposition terms35 which refer not to a
directly observable characteristic, but rather to a disposition on the
part of some physical objects to display specific reactions under
specifiable circumstances. The definition of such terms seems to
create considerable difficulties. A famous paper by Rudolph Carnap
on "Testability and Meaning33 [5] has convinced most of his fellow
philosophers that for the introduction of such a term a somewhat
different kind of logical operation is needed, which he calls partial
definition or reduction. Following HempeFs simpler presentation the
correct way to "define33 the disposition term "magnetic33 would be as
follows [10; p. 26]:

(6.4) If a small iron object is close to x at time t then x is magnetic at t,
if and only if that object moves toward x at t.

This definition is partial for one obvious reason. If there is no way
to approach x with small iron objects, e.g., x is at the bottom of a lake,
we could not determine whether it is magnetic or not. Hempel
further states [10, p. 27]:

The indeterminacy in the meaning of a term introduced by a reduction
sentence may be decreased by laying down additional reduction sentences for
it which refer to different test conditions. Thus, e.g., if the concept of electric
current had been introduced previously, (6.4) might be supplemented by the
additional reduction sentence:

(6.5) If x moves through a closed wire loop at £, then x is magnetic at t if
and only if an electric current flows in the loop at t.

Continuing this trend of thought, we find to our pleasant surprise
that the modern logician is disclosing a practice of the natural sciences,

Latent Structure Analysis 485

which, was considered to be embarrassing by many social scientists.
That is, they define important concepts as "intervening variables5' or
underlying constructs by reference to a series of test situations, which
all have to be used together.

A historian of science might one day try to prove that this em-
phasis on the special nature of disposition concepts results from the
growing importance of the behavioral sciences. It is not without inter-
est that the psychological term "disposition" is here introduced into
the epistemology of the natural sciences. The connection with the
problem of introspection is explicitly referred to in Carnap's original
paper. During an autobiographical remark on how he developed
his notion of reduction, he says [5]:

The members of our [Viennese] Circle did not wish in former times to
include into our scientific language a sentence corresponding to the English
sentence S: "This stone is not thinking about Vienna." But at present I
should prefer to construct the scientific language in such a way that it con-
tains a sentence corresponding to S.

The formal analysis of the procedure is, of course, independent of
its history and its terminology. The question is whether it really covers
the research procedure with which we are concerned here. To decide
this, we must add two more elements in HempePs exposition. First is
his distinction between the empirical and the theoretical import of
concept formation [10, p. 46, italics ours]:

In the theoretically advanced stages of science these two aspects of con-
cept formation are inseparably connected; for, as we saw, the interpretation
of a system of constructs presupposes a network of theoretical statements
in which those constructs occur. In the initial stages of research, however , which are
characterized by a largely observational vocabulary and by a low level of generalization^ it
is possible to separate the questions of empirical and of systematic import; and to do so
explicitly may be helpful for a clarification of some rather important methodological issues.

This has immediate bearing on the enterprise in which we are
engaged here. Indeed we shall concentrate on certain measurement
procedures to clarify how we create "underlying33 concepts like traits,
attitudes, group characteristics, etc.: their role is to summarize a
variety of empirical observations and to store them, one might say, for
systematic use in a "theory53 which we hope will one day develop.

No one can seriously deny that most of the social sciences are in
what Hempel refers to here as the "pre-theoretical stage of research."
On this point, then, the Carnap explication of disposition concepts is
fully transferable to our problem area.

On another point, however, we must look for an additional devel-
opment. Hempel points out what is implied when we use a variety of

486 PAUL F. LAZARSFELD

reduction sentences. Let us go back to the example of magnetism,

where attracting metal and inducing currents are used as two test
situations [103 italics ours]:

But, since the two conditions are not exhaustive of all logical possibilities,
the meaning of the word is still unspecified for many conceivable cases. On
the other hand, the test conditions clearly are not logically exclusive; both
may be satisfied by one and the same object; and for objects of this kind the
two sentences imply a specific assertion, namely: Any physical object which is
near some small iron body and moves through a closed wire loop will generate
a current in the loop if and only if it attracts the iron body. But this statement
surely is not just a stipulation concerning the use of a new term — in fact, it
does not contain the new term "magnetic,55 at all; rather, it expresses an
empirical law. Hence ^ while a single reduction sentence may be viewed simply as laying
down a notational convention for the use of the term it introduces , this is no longer possible
for a set of two or more reduction sentences concerning the same term, because such a set im-
plies, as a rule, certain statements which have the character of empirical laws.

The reader who has followed our examples in the previous sections
will have noticed that there the reduction sentences are different in one
respect. A "magnetic personality" is one which is likely to attract other
people, which is likely to induce in them currents of enthusiasm. As
we have pointed out, the items of observation are linked to the con-
cepts to be defined by probability relations. One other logician has
seen this point very clearly.

In a short paper on "Definition and Specification of Meaning,55
A. Kaplan moves on from Garnap's partial definition. He recapitulates
the position in the following words [12]:

Whenever a term is introduced into a context of inquiry . . . situations
. . . are described in which the term may be applied. Any such description
may be called an indicator for the term. But . . . indicators assign to the
application of the term under the described conditions, not a logical certainty
but only a specified weight. Thus failure to interbreed is an indicator for
distinctness of species; but that two animals do in fact interbreed does not
logically entail that they belong to the same species but only adds some weight
to the assumption.

Kaplan draws his examples from biology and occasionally from
one of the social sciences. The importance of his analysis is his clear
recognition that the relation between the indicators and the concept to
be specified does not need to have the rigid relationship implied in the
original Carnap formulation. In short, says Kaplan, "What is sug-
gested here is that indicators be formulated in terms of some type of
probable implication.53 He is also aware of an important consequence
of this more general approach to our problem: if we have two test

Latent Structure Analysis 487

situations It Is not necessary that their outcome be related by a rigid
law. To turn once more to the example of magnetism, It is now suffi-
cient to say that attracting Iron objects and inducing electric current
are correlated, that they frequently occur together but not neces-
sarily always. To this point we shall return once more.

Whereas Hempel stressed that such concept formations are char-
acteristic of any early stage of science, Kaplan stresses the fact that
they facilitate flexibility of thinking and therefore leave the road open
for new developments. In Kaplan's formulation [12]:

We begin with Indicators In terms of which the Initial application
of context can be confirmed. As the context of application grows, the specified
meaning grows — and changes — with It. The stipulation of new indicators
affects the weight of the old ones, while they in turn limit the range of choice
in the stipulation. The adequacy of a particular indicator is not judged by Its
accordance with a predetermined concept; the new and old indicators are
appraised conjointly.

Thus something which seemed to be an embarrassing shortcoming
of social science concepts, such as IQ, Introversion, or cohesion, be-
comes the common property of a large group of concept formations in
all sciences.2 In all such cases we must decide what items should be
included in the base of observations from which intervening variables
of any kind are inferred. The explication of disposition concepts thus
certainly covers all the elements we are concerned with: the use of
indicators to place people correctly into an "underlying" order re-
quired by a more abstract conceptualization, the somewhat fluid
choice of these indicators, their probability relation to the intended
"ordering," the consequent fact that they will not all point hi the same
direction and that, therefore, they have to be combined into a kind of
"index" or "measurement35 which represents the best inference which
can be made from the manifold of our empirical observations. But the
formulation of the logicians is so general that it does not lead directly
to concrete research operations. If they are our goal, one more transla-
tion has to be attempted. The notion of "property space" seems to
serve this purpose best.

The property space. The term "space" has had an interesting
biography. Originally it was used to connote the direct experience
people had when they located things in their surroundings. Then it was
seen that the points in a space could be described in algebraic terms.
Now everyone is acquainted with the notion of "coordinates." Start-

2 Carnap also stresses that often "We wish to determine the meaning of a term at
the present time for some cases only, leaving its further determination for other cases to
decisions which we intend to make step by step, on the basis of empirical knowledge
which we expect to obtain in the future" [5].

488

PAUL F. LAZARSFELD

ing, say, with the corner of a room, any other location in this room
can be indicated by saying how high up it is from the floor and how far
it is from the two walls which meet at the original corner. To each
point corresponds a triplet of distances. This leads to the extension of
the notion of dimension. Although the points in the room require three
data for their location, on a blackboard we can work with only two
coordinates — which is identical with saying that the face of the black-
board, or any other plane, has two dimensions. Correspondingly
four-dimensional sets become easy to grasp. The best known is the
space-time continuum: a bug in a room can be characterized by the
point at which it rests and the amount of time it has been there.

There developed finally an inversion of terminology. Whenever
a set of objects is characterized by a multiple of data one would talk of
them in terms of points in a space. This space would have as many
dimensions as there are data needed to characterize each of the objects
under consideration. The advantage of this terminology is that it
brings out formal similarities between materials which would be over-
looked because we habitually give them different representation. Take
as an example two students who were given three tests, language (L),
social science (S), and natural science (N). Assume their test profiles
look as follows:

Student A
StudenlB

L S N

FIG. 2. A test profile of two students.

Now the test scores are triplets of data and therefore can be con-
sidered coordinates in a three-dimensional space. To each test cor-
responds an axis and the two students thus become two points.

t Student A

N

\

\
I
I
i
I
i
l
i
i
i

'Student B
—*•$

/

i

_ „ L J

/

J

FIG. 3. The same tests as in Fig. 2 in terms of a test space.

Latent Structure Analysis 489

This translation into space terminology has the advantage that a
number of seemingly disparate notions turn out to be clearly related.
The similarity of profiles, for example, can be expressed as the distance
between points; the famous "ideal type55 becomes a special region In
the property space, e.g., the region around the origin [4].

So far our examples have all assumed that the basic data which
characterize our objects are in some way quantified. But this is not
necessary, and with this last step the most general notion of property
space is reached. The dimensions may, for example, be rank numbers
of positions in a preestablished list. All people with Christian, middle,
and family names can be put into a three-dimensional "initial space3'
in which each dimension has 26 "classes/' the letters of the alphabet.
Therefore, a man with the parameters (4,1,3) would have the initials
D. A. G., and David .Arthur Chester and Donald Avery Casey would
belong in the same "point53 in this space. In other cases the properties
might well be dichotomies, i.e., attributes which take on two values
only. Suppose, for example, people are classified according to whether
they are male or female, native or foreign born, above or below thirty-
five years of age, residing in a city above or below 100,000 population.
This would provide a space of four dimensions, but on each of them,
objects could only have two distinguishable positions; or to put it still
another way, each of the four coordinates could take on two values
only. The whole "space" would therefore consist of

2X2X2X2 = 16

"points." This space will be of basic importance for our subsequent
discussion.

The relation of manifest to latent property space. We reach the
end of this introductory section by showing how the explication of
disposition concepts and the notion of property space merge into a
rather precise formulation of our main problem. To begin with, we
must see the close connection between definition and classification.
One should not be deceived by differences in wording. Many of the
authors we reviewed seemed to ask: what is intelligence, prudence, or
friendship? Actually these writers visualize themselves as being con-
fronted with concrete cases of "intelligence" or "prudence" and want
to know how to recognize them, how to relate them to each other, and
so on. If we could ask these writers some further questions, they would
say something like this: they want to differentiate " types" of "friend-
ship" and "love" or distinguish between "prudence" and "distrust."
From a research point of view, these are all problems of classification,
although of a special kind as will be seen in the course of our discussion.

490 PAUL F. LAZARSFELD

"Measurement55 is also a special case of classification; it is irrelevant
at this point to distinguish "measurement55 from "ordering55 and other
classificatory devices.

The reader should have no difficulty in referring to the many
examples given in the preceding pages, and in verifying that the terms
mentioned, such as "traits,55 "intervening variables,55 "disposition
concepts,55 etc., are really special cases of classificatory character-
istics. They have one thing in common: they are intended character-
istics; that is, they are ways in which we want to organize a set of
objects under investigation. This locating of "objects55 (individuals,
groups, social relationships) cannot be done directly in the cases we
have discussed. We are dealing with latent characteristics, in the sense
that their parameters must somehow be derived from manifest observa-
tions. The terms manifest and latent have no connotation here beyond
the distinction between data directly accessible to the investigator
(manifest) and parameters which in some way must be inferred from
the manifest data (latent) .

The matter can be reformulated in the following way. Empirical
observations locate our objects in a manifest property space. But
this is not what we are really interested in. We want to know their
location in a latent property space. Our problem is to infer this latent space
from the manifest data. This reformulation of the relation between con-
cept formation and classification by indicators has a number of ad-
vantages. One of them deserves special attention.

In any empirical classification guided by conceptual considera-
tions we try to overcome the accidental elements inherent in the use
of indicators. Suppose we want to order people according to how they
feel about the role of government in economic affairs. We might ask
them a series of questions as to public ownership of railroads, mines,
banks, etc. It is reasonable to assume that the more someone favors
laissez faire the fewer of these items he will answer pro public owner-
ship. Still we know that many individual idiosyncrasies will creep into
the answers. A strong laissez faire person has just read about a mine
accident and under this impact he gives a pro public ownership re-
sponse to the mine item; a strong interventionist happens to know a very
fine bank president and therefore excludes the bank item from his list
of pro responses. In the manifest property space we are at the mercy of
these vagaries. But in the latent space, as we shall see, we can take
them into account and thus achieve a more "purified55 classification.

We are now ready to turn to the one question which has still been
left unanswered: how is the probability relation between the observed
indicators and the intended classification established? How do we
move from the manifest to the latent property space?

Latent Structure Analysis 491

n. THE LOGICAL FOUNDATION OF LATENT STRUCTURE ANALYSIS:
A SYNOPSIS OF THE MAIN ISSUES

Inferential classifications with the help of a set of indicators are
nothing new in the world of science. A doctor who uses a series of tests
to see whether a patient has tuberculosis, a psychoanalyst who uses
free associations to retrace a childhood experience; a chemist who
observes various reactions to identify the nature of some substance —
all use what might be called diagnostic procedures. They know, or
believe they know, laws and regularities which link their manifest indi-
cators with their latent space. Their diagnosis applies previous knowl-
edge to a specific new case.

Some initial clarifications. But there exists a second type of pro-
cedure where, so to say, the acquisition of general knowledge and its
application to a specific case are performed simultaneously. This hap-
pens if the starting point of an investigation is a statistical one and if
our attention is mainly focused on the covariation of indicators in a
large number of cases. The present section is devoted to a clarification
of this idea. It will help if we sketch the course of the following discus-
sion by raising a number of questions and offering some preliminary
answers.

In the previous section we tried to describe the intellectual climate
which led to the general idea of latent structure analysis. Now it is
necessary to describe its elements more precisely. We shall first list
them and then discuss their ramifications in some detail.

1 . With what kinds of manifest material shall we deal? They are
qualitative, but to further simplify matters, they will be dichotomies
through most of this report. Thus our examples will be "yes" or "no"
answers to an observation. Does a man agree or disagree with a state-
ment? Is he native or foreign born? Is a city above or below the na-
tional suicide rate?

We shall call any piece of such information an "item.55 In each
case we shall have an item list in which items are numbered in an
arbitrary but fixed way. The number of items in this list coincides
with the dimensionality of the manifest property space.

One alternative of each dichotomy will arbitrarily be called posi-
tive (+), the other negative ( — ). Often a judicious use of these
designations will help in intuitively grasping the material as a whole.
Each object in our study will be characterized by a response pattern of
the following kind:

Response to item no.
1 2 3 ... m

492 PAUL F. LAZARSFELD

The term response pattern is taken from questionnaire practice but is
used here in a metaphorical sense. The items might all be derived from
observation, e.g., the behavior of a person in various situations, and
the objects might be collectivities and not individuals.

2. A whole group of "respondents55 will be characterized by their
response frequencies: they are the proportion of the group who answer
each item i affirmatively (pi), two items (pij), three items (/>»•#) 9 etc. A
barred index will be used to indicate a negative response. Thus p-&% is
the proportion of people who give an affirmative answer to item 1 and
3 and a negative answer to item 2. The whole set of these response
frequencies is called a "dichotomous system55 and its nature is very
important for a more detailed study of latent structure analysis.
For our present purpose, acquaintance with the symbolism will suffice.

3. How are we to represent a latent space? In the traditional way
by a system of coordinate axes, i.e., a so-called cartesian frame of
reference. An example of a three-dimensional frame of reference was
given in Fig. 3. Actually in this section we shall restrict ourselves to
one-dimensional latent spaces to facilitate exposition. But in a later
section we shall see that assessing the number of dimensions in the
latent space is possible. A one-dimensional space is, of course, a straight
line. We shall often call it a latent continuum.

4. What is the relation between the manifest items and the latent
continuum? It involves probabilities. Thus we shall assume that there
exists a curve — preferably a mathematically simple one — which
relates to each point of the latent continuum a specific probability
that a given item has a positive response. Suppose that our latent
continuum deals with socioeconomic status, and three of our items are

ownership of a yacht, presence of

Running warm water ^ running warm water, and presence
of two living rooms in the house of
the respondent. Common sense
would let us suspect that the

corresponding probability rela-
Socio-economic status — ^ tion ^ be somewhat like Fig_ 4.

FIG. 4. Probability relations between the rpi , . . . . ., .

, , , .- J . , . . The srraph intimates that very

intended classification by socioeconomic -11 -, . -, /

status and the observed frequency of quickly as we go up the social scale
three indicators. almost every family will have run-

ning warm water; only the upper

crust is likely to own yachts, while the probability of more space in
addition to bedrooms increases fairly proportionally with socio-
economic status. The truth of these surmises is not relevant here.
What matters is the way they are expressed through Fig. 4. Of course
we have not yet stated precisely what we mean by probability, and we

Latent Structure Analysis 493

are still in the dark as to how we would define and know a respondent's
socioeconomic status. These two points soon will be taken up in
considerable detail.

The curves of probabilities in Fig. 4 we shall call trace lines; they
trace the probability for an item as a c 'respondent53 moves along the
latent continuum. If the latter is a two-dimensional one then the
probabilities form a trace surface. In full generality we shall talk of
latent traces. Notice that a trace line is defined for each item separately.
Later a crucial problem will be what to think about the probability of
joint responses to several items occurring simultaneously at each point
of the latent space.

5. How are we to understand the term "probability" used all
through the preceding pages? A traditional example of how to look
at probability is as follows: we take a sample of people aged fifty
and find out how many die within the next year; we compute the
proportion of people who died between the ages of fifty and fifty-one.
Then we generalize this ratio and say that it is the probability of dying
at the age of fifty within one year. This operation can be refined as far
as we want to go. We might say, for instance, that the probability of
business executives5 dying within a year at the age of fifty is greater
than the corresponding probability for office clerks. The class for
which such probabilities are computed and then generalized is usually
called the reference class of the probability [21].

This same procedure, however, can be used in still another way.
Suppose we ask an individual, Mr. Brown, repeatedly whether he is in
favor of the United Nations; suppose further that after each question
we "wash his brains'5 and ask him the same question again. Because
Mr. Brown is not certain as to how he feels about the United Nations,
he will sometimes give a favorable and sometimes an unfavorable
answer. Having gone through this procedure many times, we then
compute the proportion of times Mr. Brown was in favor of the United
Nations. This we could also call the probability of Mr. Brown's being in
favor of the United Nations. But now the reference class is not many
Mr. Browns having been asked this question once, but one Mr. Brown
having been asked the question many times.

There is one interesting consequence of this version of the probabil-
ity notion. A specific Mr. Brown, for instance, might feel that it is his
duty to be in favor of the United Nations. Therefore, if he is asked a
question when he is sober, his probability — or, if you please, his
propensity — to be in favor of the United Nations might be rather high.
Under the influence of alcohol, however, his hostility to the interna-
tional organization might come out. Therefore, his probability under
the influence of alcohol could be different than his probability if he

494 PAUL F. LAZARSFELD

were sober. This is an Idea which is often used in the parlance of daily
life. A man can drive "60 miles an hour" and at the next moment if a
policeman is around, slow down to CC40 miles an hour." What we call
miles per hour is not what a man actually drives within an hour, but
what he would drive if certain specified conditions were to prevail.
Thus, we have a concept of probability which can apply to a single
individual; furthermore, this probability or propensity itself can be
different under various conditions.

6. How can we know trace lines? This is, of course, the central
problem of latent structure analysis, and the third section of this report
is given to developing an appropriate answer. The remainder of the
present section will provide some preparation, considerations, and
examples.

The whole configuration of trace lines for all items and the loca-
tion of each object in the latent space is called the latent structure. It
is a typical example of what is often called a "mathematical model,'5 a
construct which is derived from actual data together with certain
general reflections on the purpose these data serve. In our case the
situation is as follows: from our manifest data we actually know the fre-
quencies in which the various response patterns occur in a given popu-
lation; what we want to know are the latent parameters of the model, the
coefficients which characterize the latent traces, and the distribution
of the population within the latent space. We therefore need equations
which link the manifest frequencies to the latent parameters. From
these so-called accounting equations we then can compute all the elements
in the model. The name given to these equations is meant to indicate
that with the knowledge of the full latent structure, we can account for
everything known about the manifest data.

In order to clarify this basic idea, it is best to discuss in some detail
two empirical operations with which most research students are well
acquainted. By a slight extrapolation they become basic elements of
latent structure analysis. The situation is somewhat similar to what
was just mentioned about the concept of probability. Probabilities are
formal extrapolations from the empirical notion of relative frequencies.
In our case we are referring to item analysis which forms the basis of
the notion of trace lines, and to the "explanation" of statistical rela-
tions which becomes the basis for the accounting equations.

We turn first to item analysis and trace lines.

Item analysis and item curves. Every graduate student who
takes a course in applied psychology knows about item analysis. If he
wants to develop an attitude or a performance test he knows that he
should proceed in the following way. He is permitted to start with
many questionnaire items which he hopes will be indicative of what in

Latent Structure Analysis

495

the end he wants to "measure." But then he Is supposed to distinguish
between good items and bad items. This he is taught to do in the fol-
lowing way: he forms a ccraw" score by adding up for each respondent
the number of items which are answered in the "correct" way. Then
he plots each single item against this raw score. (TMs we shall call
the item curves.} The items which have a high association with the raw
score are acceptable. The items which have a low association are
considered inappropriate and should be eliminated. We will now give
an example of such an item analysis, but we will refine it in two ways:

1 . We shall plot two items against the raw score.

2. We shall not only plot each item separately, but we shall investi-
gate how the association between the two items is related to the raw score.

Item analysis applied simultaneously to more than one item.
Our material comes from a public relations study where 560 re-
spondents were asked questions regarding their attitudes to the oil
industry. Do oil companies treat their workers fairly; do they make
too much profit; are they wasteful of our natural resources, etc.? To
each question the respondent could give one of five answers, which
ranged from firmly favorable to firmly unfavorable (from the oil
industry's point of view). There were ten questions in all and eight of
them were combined into an arbitrary score in the following way: a
firmly favorable answer was given a weight of 4 and so on down to the
firmly unfavorable, which got a weight of 0. Then all the weights were
added so that a respondent's general attitude score could range from
0 to 32. This score was used as the "outside continuum" or base
variable. Against it the probabilities (proportion) of answers to the
remaining two questions were plotted. These two were:

Item 1. Do the big oil companies control too much of the oil
business?

Item 2. Is the oil industry wasteful of our natural resources?

TABLE 1. THE INTERRELATION OF Two TEST ITEMS FOR FIVE SUBCLASSES OF
RESPONDENTS CLASSIFIED ACCORDING TO A GENERAL ATTITUDE SCORE /

DERIVED FROM EIGHT OTHER QUESTIONS*
Item 2 Item 2 Item 2 Item 2 Item 2

- +

3

14

17- +

15

27

42- +

35

26

61 - +

40

18

58 - +

67

17

84

E

E

£

E

£

<D

a>

o

03

30

59

89 ±:-

32

61

93 2=-

34

29

63 ±=-

13

21

34 --

12

7

19

Total 33 73 106 47 88 135 69 55 124 53 39 92 79 24 103

General

attitude Oto16 17to20 21 to 23 24to26 27 to 32

score

* A response favorable to the oil industry is indicated by a + sign.

496

PAUL F. LAZARSFELD

A positive reply (from the industry point of view) was one in which
a respondent expressed at least some disagreement (score class 3 and
4). The joint positive response required such disagreement with both
items. In Table 1 we now have the data for the item curves for each
item alone and for the joint responses. The vertical marginals of the
five partial fourfold tables indicate the item curve of item 1 . In the
lowest general attitude group, 17 out of 106, or 16 per cent, give an
affirmative response. In the group farthest to the right, 84 out of 1033
or 84 per cent, do so. The horizontal marginals indicate the item curve
for the second item; the corresponding figures are 31 per cent and 79
per cent. Thus item 1 (concern with economic control) is more expres-

90
80
70
60
50
40
30
20
10

land 2

B CD

FIG. 5. Item curves corresponding to Table 1.

sive of the "underlying53 classification than item 2 (concern with
natural resources). Figure 5 shows the item curves for items 1 and 2
and adds a third: the proportion of people who give a positive answer
to both items. These proportions are based on the left upper corner
figures of each of the five fourfold tables in Table 1. Note that the
item curve for both items is more concave (seen from the top) than
either of the curves for items 1 and 2 separately.

The items in our attitude tests have obviously been selected by the
investigator as indicators of an underlying continuum according to the
reasoning discussed in our historical section. The item curves are
a crude representation of the relation between these indicators and
the intended classification. But what about the relation among the
indicators? The reader will remember that we came to the general ex-
pectation that indicators will be statistically related to each other be-
cause they have their links with the underlying continuum in common.
In terms of probability notions we can now put it this way: the proba-
bility of joint occurrence p^ will not be pipt — the chance result of two

160

102

262

298
560

121

177

281 279

Latent Structure Analysis 497

Independent probabilities — but, rather, p^ will be greater than/?^2. In
the empirical data we shall expect a positive association in the fourfold
table, which cross-tabulates the reply to two items. This turns out to
be the case in our public relations example, as can be seen from
Table 2.

Table 2 is obtained by adding the five partial tables in Table 1 .

TABLE 2. THE INTERRELATION OF ITEMS 1 AND 2 OF TABLE 1,
FOR ALL 560 RESPONDENTS
Item 2

Item 1

Here/?12 = 160/560 = .29, whereas/?! -j&2 is only (.47) (.50) = .24.
Still, using rather informal language, we can say that the responses to
two items of a test show positive relations because they were chosen as
indicators of an underlying property. But this argument can be turned
around. // a class of people are alike in an underlying property, then the indi-
cators of this property should not be statistically related in this class. In our
example we can submit this idea to a crude test. Our general score was
supposed to be a crude measure of the general attitude of the respond-
ent to the oil industry. By dividing the respondents into five classes, as
in Table 1, we get groups of people who among themselves have a
rather similar attitude.

In general our expectation is borne out. We now have five partial
fourfold tables: the association is negative in one, practically zero in
two, and positive in two. The five tables can be looked upon as chance
variations from an association which is actually zero. Extrapolating
the results of Table 1, we can say that if by an appropriate score, the
underlying property of a population is kept constant, then the indi-
cators of the property are statistically unrelated.3

3 Whether the association or correlation between items 1 and 2 in such sub-
classifications can be considered a result of chance can be tested by x2 procedures.
It should be desirable to obtain more such examples, because the present one points
to an interesting possibility. The five associations go uniformly from negative to
positive as we move from low to high general scores. If this turns out to be the case
in other tests, we would be confronted with a result in test psychology which deserves
further investigation and interpretation. For our present purpose this matter is
irrelevant, because we use our concrete data only to lead up to an axiomatic
idealization.

498

PAUL F. LAZARSFELD

The essence of our example Is this. On the basis of their raw scores,
we divided our respondents into five classes. Within these classes raw
scores are relatively similar, and this similarity is understood to
correspond to the similarity of individuals within a class in their
general attitude toward the oil industry. Just as individuals within
a class have a similar attitude, individuals in different classes have
dissimilar attitudes. This dissimilarity manifests itself in the differences
between probabilities of affirmative responses in the different classes.
Within each class the probabilities for all people are the same for any
one item; of course, different items will generally have different proba-
bilities of affirmative response within a given class. Considering the
responses to a single item by individuals in one class, we still find a
mixture of positive and negative responses. After all, the class does not
determine the response; it only determines the probability of each
response. The variability of response is supposed to stem from acci-
dental elements. Quite irrespective of their attitude toward a specific
industry, some people happen to be more concerned with the preserva-
tion of natural resources, others are more worried about the growth
of economic monopoly. Biographical and other reasons might ac-
count for such a difference; in any case, these idiosyncratic elements
are assumed to be unrelated to each other. Within a class which is
homogeneous in regard to its basic attitude, the answers to specific
items are assumed to be unrelated. This was not quite the case in our
concrete example but was enough so that an extrapolation seems
indicated. We shall define a homogeneous class as one in which this statistical
independence of indicators prevails.

This leads us to investigate the characteristics of a group of re-
spondents which can be considered a mixture of subgroups where,
within the subgroups, a set of indicators are statistically independent
of each other.

The "mixture" phenomenon and its role in the explanation of
statistical relations. We start with a simple case. Suppose that in
each of three groups two items are statistically independent. Table 3
should be looked upon as an idealization (and simplification) of Table
1. We "mix53 these three groups and form a new one by adding box by
box the corresponding numbers on the left side of Table 3. On the
right side we now find an association between the two items, which

TABLE 3

12 24 36

18 18

54

54 54 108

Latent Structure Analysis

499

did not exist in the three partial tables to the left. Where does this
association come from? We understand this best if we look at the
margins of the three left-side parts of Table 3. In Class I the probabil-
ity of a positive response is much lower on both attributes than in Class
III. If in the combined population we select successive respondents,
they will sometimes come from Class I and sometimes from Class III.
In the former case, they will be more likely to give negative responses
on both attributes; in the latter case, both responses are more likely
to be positive. The statistical association between the two items in the total
population is thus accounted for by the fact that each attribute by itself is posi-
tively related to the general attitude of the respondents which distinguishes the
three subclasses. This was, of course, also the case in our previous exam-
ple of the public relations study of the oil industry. Table 1 shows that,
for all the items, the probability of a positive response increases with
the general attitude score. The positive association between the two
items in Table 2 is therefore accounted for by the fact that they are
both indicators of an underlying attitude or, more precisely, the
probability of positive responses is positively related to the general
attitude score. The resulting association in Table 3 is noticeable but
not very strong. Therefore we increase the marginal differences be-
tween the three homogeneous classes and mix again. Now the resulting

TABLE 4

30 36

18 18 36

54 54 108

association is much more marked. How is this finding to be explained
in the light of the previous discussion? It will be remembered that the
marginals in the partial subtables correspond to the "item curves'3 of
the two items. In Table 4 they are clearly much steeper than in Table
3. This means that now the two items have a much stronger relation to
the underlying continuum than before. As a result the interrelation
between the two indicators on the right side of Table 4 is much
stronger than in Table 3.4

So far our emphasis has been on the mixing of homogeneous sub-
groups and the resulting associations between indicators. But Tables 1,

4 The reader should satisfy himself that many other combinations could occur.
Suppose, e.g., that we made Class III in Tables 3 and 4 much larger than Class I.
Then, in the resulting fourfold table, both items and their joint occurrence would
show higher frequencies.

500 PAUL F. LAZARSFELD

3, and 4 can also be read in the opposite direction, from right to left.
We then start with an existing association between indicators; we
"unmix" the population under study and end by showing the homo-
geneous subgroups in which the associations disappear. Actually, this is
always done if, in empirical research, a statistical finding is to be
explained. We want to remind the reader of the three major types of
such explanations.

Type I. A good example is available from political research. In a
presidential election educated people vote more frequently than the
uneducated. We can classify people into three groups, however,
according to their interest in elections. Then on each interest level we
can set up a fourfold table between voting and education (graduation
from high school forming the point where higher education begins).
We then find that with each increase in interest the proportion of
voters increases, as well as the proportion of people having higher
education. Within interest groups, however, there is practically no
relation between education and voting. Interest, therefore, accounts
for the original relation in terms of what is usually called an "inter-
vening53 variable. The whole structure can be represented by the
following scheme where arrows stand for a vague idea of causation.

Education — * Interest — > Voting

The original two variables are underscored. Their association is
interpreted through the role of "interest." The interpretation is tested
by showing that the original association disappears within subgroups
which are homogeneous in regard to interest.5

Type 2. The second major type of accounting is usually known
as the controlling of spurious factors.

Examples are almost proverbial: fires where many fire engines
come out cause more damage; does this mean that fire engines are
dangerous? Obviously not. Large fires bring out many engines and
cause much damage. The arrow scheme corresponding to this case
would be as follows:

Amount of ^ Damage

equipment

Size of fire

5 Cf. [15]. There a characteristic counterexample is included. Men vote more fre-
quently than women. This cannot be accounted for by interest. Even within the same
interest group men vote more than women.

Latent Structure Analysis 501

If the size of the fire Is sckept constant" there would be no positive
relation between equipment and damage.

In both types of accounting the statistical test is the same, and it is
the one which we have carefully analyzed above. The ultimate rela-
tion between two attributes is owing to the fact that they are both
related to a third property; once this property is kept constant the
original relation disappears. The difference between type 1 and type 2
lies in the sequence of variables involved. In both cases we start with
an association between two factors: education and voting in the first
example, equipment and damage in the second. But in type 1 the
explanatory factor intervenes between the two original variables;
whereas in type 2 it antecedes the damage as well as the number of
engines the fire brings out.

Type 3. The third type of accounting is usually less discussed
because its outcome seems so obvious from a substantive point of view.
Still it is the most important one for the present purpose, and all our
initial examples belong here. When we deal with indicators of a sup-
posed underlying property, there exists no necessary time relation
between the intended classification and its overt manifestations.
Rather the relation here is one of generality and specificity. Still, the
test of whether we are really dealing with appropriate indicators is
the same as before; we want to know whether the underlying property does
account for the interrelation between the manifest indicators. We would look
for a way to classify people according to the underlying characteristic
and assume that if this is held constant no further statistical relation
should exist between the various indicators.

The most obvious way to make this test is to see whether people
who are alike on the majority of the indicators show any appreciable
relation between the remaining ones. This we did with our oil study
example. Another approach would be to use a rating scale. For
instance, people could be asked to rate their political interest on a
scale from 1 to 10. Then they could be divided into fairly homogeneous
classes according to this self-rating. If we then have an itemized inter-
est test, we could raise this question: does the self-rating account for
the interrelation between the test items? The answer would be in the
positive if, on each level of self-rating, the items were not statistically
associated. The arrow scheme corresponding to that of the previous
example would be :

Item 1 Item 2

self-rating
on interest
Item 3 *^^

502 PAUL F. LAZARSFELD

It is hoped that we have given enough examples so that the reader
has a clear picture of the following two related facts:

1 . There exists a uniform operation of accounting for an empirical
relation between two properties. It consists of studying this relation for
subclasses of the original population, these subclasses being formed by
the introduction of additional properties. The substantive nature of
these "accounting properties55 and their relation with the original
data make for the main types of accounting procedures as they occur in
the practice of research.

2. These various accounting procedures are all in fact inversions of
the "mixture phenomenon" described above. In mixtures of homo-
geneous groups, indicators show statistical associations: they are due to
the covariation of the indicators between these subclasses. Inversely,
associations between indicators in empirical populations can be
accounted for by dividing them into homogeneous subgroups; the
variables along which this " unmixing' 5 can be done "explain" the
statistical associations originally found.

We can now relate all these considerations to latent structure
analysis.

The accounting equations and the principle of local independ-
ence. We are prepared to answer the question which we first raised
at the end of the historical survey in the first section. There we came to
the conclusion that the problem of disposition concepts boiled down to
the task of relating a manifest to a latent property space. The manifest space
was given by the observed properties of our objects, which for our
present purpose, we have reduced to dichotomies. We came to use the
word "item" for an indicator and the term "response" for its observed
presence or absence. A "response pattern" was a point in such a
dichotomous property space.

The latent space corresponds to our intended classification, which,
as we saw, was variously called in the literature "an underlying
characteristic," "a trait," "a disposition," etc. It was not necessary
to assume anything about this latent space; the number of its dimen-
sions was not specified, nor did they need to be of any particular
mathematical form, dichotomous, continuous, or whatever. But when
we came to this point of our discussion we left it undecided how we
would achieve this latent classification, in view of the fact that we
have only manifest observations available. What solution does latent
structure analysis propose for this problem?

It defines the latent space as that classification which accounts for
the statistical interrelations between the manifest observed indicators.
It is the classification which "unmixes" a given population into homo-
geneous subgroups. The many consequences of this definition must
now be spelled out.

Latent Structure Analysis 503

The latent space Is not known in advance but is defined by its ac-
counting role. When we discussed examples of Item curves, \ve
first divided people by some index (e.g., a rating on the number of
positive responses) into fairly homogeneous classes and then studied
empirically howT the Indicators were related in the various subclasses;
Table 1 was a typical example. But in latent structure analysis we do not
have an empirically provided general classification against which the occurrence
frequencies or probabilities of the different items can be plotted. The underlying
classification is derived from the satistical behavior of the indicators themselves.
The sequence of affirmative response proportions for a given item over
all homogeneous cc latent3 ' classes becomes its item curve; it Is now
called a trace line to stress that it is not directly given but derived from
empirical data.

Accounting equations. Let us assume that we have c homogeneous
classes and n items. Let us assume further that these homogeneous
classes are ordered in some way. (In Table 3, for instance c = 3; and
the three classes are ordered from left to right.) Let us now focus on
two specific items, say, the two items in the scheme just mentioned.
The response frequencies of the two items in the composite population
(exemplified by the right side of Table 3) can be derived from the
following equations:

pi =

This is in algebraic form the box-by-box summation we have carried
out in all our mixing examples. The superscript in p-f shows from
what class x the "latent probabilities" have been taken; vx is the
proportion of the whole population in class x. Suppose now the
joint frequencies on the left were given and the task consisted in
computing the response frequencies in the homogeneous subclasses.
We could not solve it because there are fewer equations than un-
knowns. Obviously, however, we could add more equations by adding
more items. Not only would that give us more equations of the type
just mentioned, but it would also add an additional type: we could
now set up equations for higher-order frequencies, for instance,

pin =

504 PAUL F. LAZARSFELD

As we will see later, in general there always occurs a point where
we have enough equations to solve the whole problem. Equations of
this type are called accounting equations because they permit us to
derive parameters of the latent structure from the manifest data.
These accounting equations are, in a way, the mathematical summary
of everything we have said so far. Let us review, therefore, how they
are related to the different elements of our discussion.

First, they formalize algebraically the diagnostic procedure
by which we make the inference from the manifest data to the latent
position of a respondent. An indicator or test item is introduced be-
cause we have a more or less vague idea how it is likely to be related
to what we want to find out about each of our respondents. We as-
sume that if we could by some manipulation put people into various
positions of this intended classification, their response probability would
vary according to this general image. The latent probabilities give
precise expression to the relation between the latent and the manifest
space. They are tantamount to a diagnosis for any manifest response
pattern which might be empirically observed.

But, of course, we cannot move respondents into various positions;
we observe each respondent only once. Instead, we have a variety of
respondents whom we assume — in the spirit of the whole model — to
be actually at different places in the latent structure. As a matter of
fact, we make an even stronger assumption: for the purpose of our
model, we assume that all our respondents are alike but for one fact,
that they are different in regard to the latent property. This is the
second element in the whole analysis.

A third element is the following trend of reasoning. Even if we
knew where a respondent belongs in the latent space, we would have to
make him respond repeatedly to each item so that we could ascertain
empirically his response probability. But again we must remember that
each respondent is observed only once on each of the items. This diffi-
culty is surmounted by the idea that in an empirical population we are
most likely to have many people who are at the same point in the
latent space. Now we consider such respondents to be, for our purpose,
identical; therefore, the proportion of affirmative answers in such a
homogeneous group can be taken to be the same as we would have
obtained if we had observed one member of each of these groups
repeatedly.

Let us get this series of constructions clearly in our minds by visual-
izing the process in reverse order. We could get all our trace lines by
the following procedure:

1 . We imbue one respondent, by some kind of manipulation, with
various amounts of the latent property.

Latent Structure Analysis 505

2. At each point, we make him respond to each Item repeatedly
with "brainwashing55 inserted between any two trials. This would give
us, at each point of the latent space, the probability of an affirmative
response for our "typical55 subjects.

3. The totality of these probabilities, attached to each point of the
latent space, would be the trace of an item — in the one-dimensional
case, the trace line.

Now the steps (1) and (2) are replaced by the fact that at each
point of the latent space we have many respondents; \ve substitute
their response frequencies for the probabilities we are looking for.
But remember that even this is a fiction. Although we are convinced
that our whole population can be subclassified into such homogeneous
groups, when we deal with concrete respondents \ve do not know at
what point of the latent space they are. Here we take advantage of the
accounting equations just explained and developed. What wre actually
know are the response frequencies of a mixed population to a number
of items in all their combinations. Therefrom wre can compute the
response probabilities in the postulated homogeneous subclasses. As a
matter of fact, we learn, as a by-product of this computation, what
proportion of our respondents is in each of these latent classes.

But notice that even now there is one topic which w7e have not
discussed at all, namely, the single respondent. We do not know at
which point of the latent space he is located. This is a matter which
we will take up only in the next section. What we now know is the
latent structure, the proportion of people in each class and the condi-
tional probabilities of giving an affirmative response to each item in
these classes.

This whole web of assumptions and deductions can be fruitfully
divided into three sections. One has to do with rather conventional
ideas which are accepted wherever probability notions are introduced;
the idea, for instance, that for the purpose of a specific investigation
different people can be considered as alike, and that the proportion
having a property is an estimate of the probability that a single one of
them will exhibit it. Although the logical foundations of this idea are
by no means simple, we need not justify them here because of their
general acceptance in all model building.

A second group of our ideas has to do with the problem of un-
mixing: deriving the probabilities in homogeneous subclasses from the
response frequencies of a mixed population. This is straightforward
algebra and does not require any further logical foundation. Actually,
it is the most characteristic and novel aspect of latent structure analy-
sis, and much of the rest of our monograph will elaborate on it. Here
the accounting equations come in.

506 PAUL F. LAZARSFELD

Finally, we have the principle of local independence. (The term has
been suggested by Frederick Hosteller.) It covers the phase of our
discussion in which an intended classification (an underlying, inter-
vening variable) is defined as the one which divides a given population
into homogeneous subgroups. The principle of local independence
identifies the "measurement35 problem with the mixing phenomenon
or, rather, its inversion — unmixing. For this, no further foundation
can be introduced. The principle is proposed as a mathematical axiom
which formalizes the basic assumption of what we have called index
formation in the social sciences. If an investigator chooses a number of
indicators for the purpose of diagnosis, for the purpose of putting
people or social objects into an intended classification, he does assume
— knowingly or not — that the statistical relations between these
indicators are essentially owing to the fact that they are all related to
the intended latent property. For a group of people, therefore, who are
alike in regard to this latent property, all the indicators will be sta-
tistically unrelated. This principle partakes of the common character-
istics of all axioms which are introduced into a theory — and no theory
exists without at least one axiom: if, after all proper consideration,
some specific empirical data seem to contradict the axiom, then the
investigator will decide that "there is something wrong about the
data53 and will maintain the axiom. This formulation, put by purpose
in an almost paradoxical form, will be carefully amplified in subse-
quent sections.6

The idea of making the principle of local independence the nub
of index construction, even of concept formation in the social sciences,
is the central logical feature of latent structure analysis. Together
with conventional probability notions and some newly developed
but quite orthodox algebra, all procedures and all empirical findings
derive from it.

We shall now present the main steps in an actual, relatively simple
application.

HI. THE NINE STEPS OF LATENT STRUCTURE ANALYSIS

A latent structure analysis of necessity involves a certain sequence
of operations which can be cast into a schedule of nine steps.

Summary of 'the nine steps. First we must think about the form of
models which might reasonably be appropriate. This means that wre
want to consider systems of manifest and latent variables such that
their interrelations mirror the interrelations between indicators in the

6 Frederic Lord in discussing the principle of local independence has aptly stated
that it is "almost indispensable for any theory of measurement." Gf. [17].

Latent Structure Analysis 507

data, and the concept which is the real object of concern. Having
considered these questions in a general way, we must state our
assumptions In explicit mathematical form (step 1). Then we can
write the accounting equations, wiiich give the relations between
manifest and latent parameters. In the particular form which the
chosen model imposes on them (step 2). Next we must ask what con-
ditions or restrictions are put on the interrelations within the data
by the assumptions of the model (step 3). These ''conditions of re-
ducibility5' are useful in a number of ways: first, they are explicit
statements of relations which must hold among the manifest param-
eters, so that by means of simple operations on the data, and with-
out solving the accounting equations, we can determine whether the
assumed model is appropriate for the given data. Second, the condi-
tions of reducibility are useful in evaluating how closely the data are
in accord with the requirements of the model. Third, the conditions
of reducibility contribute to an understanding of the model and to
the question of the solvability of the accounting equations.

This question of the solvability of the accounting equations may
be asked more specifically in the form: given the manifest parameters,
are there a sufficient number of conditions imposed by the model to
make it possible to identify the latent parameters? (step 4) Having
answered this question, we proceed to its logical corollary: If the
equations are solvable, how does one actually solve them? (step 5)

Up to this point everything is algebra. Now the data must be
introduced, and we are forced to do some arithmetic. A "fitting
procedure" (step 6) in latent structure analysis Is usually a shuttling
back and forth between data and latent parameters — using data of
lower order to identify certain latent parameters, from these com-
puting what the data would have been if they had fitted the (partially
identified) model perfectly, then combining these "fitted data" with
higher-order manifest data to compute further latent parameters. A
fitting procedure has two goals: (a) a set of latent parameters and
(b) a set of "fitted manifest parameters" which are perfectly in agree-
ment with the demands of the model and at the same time are as
close as possible to the actual data. How close a fit was achieved re-
quires some evaluation (step 7). Two questions must be answered
here: Are the differences between the actual and the fitted parameters
small enough? Do the differences appear to be randomly distributed,
or do they fall into some pattern which suggests that a somewhat
different latent structure model would be more appropriate? If this
is the case, we must again start from scratch, except for what we
have learned by the experience. But if we are satisfied with the fit,
there is still some work to be done.

508

PAUL F. LAZARSFELD

We want to know how the respondents who give a particular
response pattern are distributed over the latent space (step 8). And
it is of interest to ask about the most likely or the most typical location
of individuals who gave a particular response pattern, and in some
way to assign a score to each respondent, or to each response pattern.
Similarly we may ask how much each item contributes to the diag-
nostic process, and perhaps we may wish to give each item a score
indicating its ability to discriminate between individuals at different
points of the latent space (step 9).

For illustrative purposes we shall, throughout our description of
the nine steps, make use of one simple latent structure model, the
so-called linear traceline model.

In order to illustrate what we have to say with concrete numerical
data, we have taken six items which were included in a questionnaire
because it was thought that they would serve as indicators for the
concept of "job satisfaction.35 These questions were answered by 876
employees of a large industrial concern. Note that with each question
is given a definition of what is considered a positive response, and the
proportion pi of all respondents giving this response.

Question

Positive response

1. "Are there any things about your job that you
particularly like?"

2. "Are there any things about your job that you
particularly dislike?"

3. "How often do you look forward with some
pleasure to your day on the job?"

4. "If someone asked you about getting a job like
yours, which of the following would you be in-
clined to do? Encourage her? Discourage her?
Neither?55

5. "Do you ever feel you would like to quit and
get a job with some other company?"

6. "Do you feel that you would like to get a
transfer from your present job to some other
kind of work in your department?"

"A lot of things"
"None" and "not many"

"Every day" and "almost

every day"
"Encourage her"

"Never"

"Seldom" and "never"

.34
.57
.62
.48

.38
.58

We are restricting our discussion to these six items for the sake of
simplicity — by means of them we can illustrate quite well what we
want to present. Indeed, when four items suffice to illustrate a point,
we shall use only four. In the actual questionnaire, however, there
were more items tapping the notion of "job satisfaction.33

Step 1: Choice and specification of the model. The first problem
to be faced is the choice of a specific model. In our example we are

Latent Structure Analysis 509

assuming that job satisfaction is a single dimension. The six questions
just quoted are the indicators for the latent continuum. In this special
case the probability of an affirmative response to each item is
proportional to the degree of "underlying satisfaction." This means
mathematically that all the trace lines are linear functions

/«(*) = *i° + a+x (I)

The two parameters of such a trace line correspond to different
elements in the content of a questionnaire Item, flj1, which is the slope
of the trace line, indicates something like the discriminating power of
the item. If the slope of the trace line is steep, then a small increase in
job satisfaction will lead to a considerable increase in the probability
of an affirmative answer. If the trace line is flat, then the answer to an
item is not very indicative of the latent variable. We will see, for
instance, that this is the case for question 4. Maybe people do not ask
for transfers even if they are not very satisfied with their jobs because
they feel that there is not much difference among the various jobs they
have a chance to get.

The coefficient af corresponds to the probability that a question is
answered affirmatively irrespective of a respondent's job satisfaction.

What about the distribution of the population over the latent
space? For the model at hand, very little need be said about it. Indeed,
we shall put no restrictions on this distribution except one which is
implied by the form of our trace lines and the fact that they represent
probabilities. Probabilities cannot assume values greater than 1 or less
than 0. But any straight line will eventually escape from these bounds,
unless it is horizontal. Consequently, we must rely on the distribution
function to make sure that whenever a trace line is larger than 1 or less
than 0, there is "nobody at home," the density function <f>(x) is 0.

The choice of the model then expresses in mathematical form
certain substantive notions which the investigator had in mind when
he collected his data. Whether his expectations are justified will only
be known in step 7.

Step 2: Accounting equations specialized for the model. We
have so far talked about the equations relating manifest and latent
variables only in general form. When we now bring the specification
made in step 1 to the accounting equations, the first-order accounting
equations for the linear traceline model come to be

*) dx

= a* 0(*) dx + fli1 *0(*) dx
= a* + ailMl (2)

510 PAUL F. LAZARSFELD

4>(x) dx = 1 by definition, since <$>(x) is a probability density
function. Mk, the kth moment of a distribution, is denned as

where the first moment is the ordinary mean of the distribution.
For the second-order accounting equations we have

dx

= a€yo + flfy Wi + aysAfa (3)

The last line Introduces a convenient shorthand notation. The ele-
ments at/* arise in the multiplication of polynomials as the coefficients
of the kth power of x. The symbolism is easy to understand: the sub-
scripts indicate which items are involved; the superscript is the power
of ;c3 or the order of the moment, to which the coefficient is attached.
These coefficients, also called "convolutions/5 make possible even
greater economies in notation for the accounting equations for higher-
order frequencies. These accounting equations are similar to Eqs. (2)
and (3)3 the integrands being, however, products of larger numbers
of linear expressions of the form (1). Thus the accounting equations
for third-order frequencies are obtained by the integration of a product
of three linear expressions, giving rise to so-called third-order con-
volutions as the coefficients of the moments:

+ cnj^-M-i + ai^Mz + aijk*Mz (4)

The reader will have no difficulty in verifying that, for instance,

The right side of (4a) is the sum of all possible products involving one
coefficient from each of the three items z, /, and A, the sum of super-
scripts in each term being equal to the superscript of the term on the
left. This may be considered a defining property of the convolutions;
but we cannot here enter a detailed discussion of this intrinsically
interesting topic [7]. Note that the third moment M% occurs in the
accounting equation for third-order frequencies. For the linear trace-
line model this is a general situation: the accounting equations for a
manifest frequency of order n involve all moments up to the rath.

If we deal with m items, we have to write 2m accounting equations.
A very important condensation can be achieved if matrix algebra is
used. Let us take as an example the second-order frequencies. The

Latent Structure Analysis

511

proportion of joint affirmative answers to Items z and j can be written

In the form [1]

, o n f 1 Mi ' f a? ^

= k-V) ,

(5)

4/7 the manifest second-order frequencies can be put in the form of a
matrix P as follows:

— pn pu . . .

p =

(6)

The accounting equations for all the manifest frequencies in the matrix
P can be combined in one matrix equation. We introduce two more
matrices. The one consists of moments of the latent distribution func-
tion, and the other consists of the coefficients of all the trace lines. The
first — the so-called moment matrix M — has already been exhibited on
the right side of Eq. (5). The matrix of the latent traceline coefficients
is defined as follows:

A =

(7)

All the accounting equations of the second order then can be written as

P = A'MA (8)

where A7 is the transpose of A.

Similar equations can be developed for higher-order frequencies.
As an example we present the accounting equations for the third-order
frequencies in matrix form. The accounting equation for the third-
order frequency p^ can be put in a form similar to Eq. (5)3 namely,

i i /i/i ^ /i/i _ i
ptih = 0*° a*1)

1

Mi

0 a*

(9)

How can third-order frequencies be combined into matrices? There
are a variety of answers to this question. For our present purpose the
best way is to consider a selection of third-order data in a stratified
matrix (10). It comes about by attaching to the entries of P — see
Eq. (6) — one additional index, the so-called stratifier. This gives us:

plmk

pmlk

(10)

512 PAUL F. LAZARSFELD

Now we want an equation like Eq. (8) which shows how the mani-
fest matrix P& looks In terms of latent parameters. This time our de-
parture is Eq. (9). It is easy to verify that

A /

Is — A

1 Ml

&K U

akl ak°
0 a*1

A (11)

The reader who is not well acquainted with the shorthand value of
matrix equations can get the gist of the story merely by inspecting
Eqs. (5) and (9). Equation (8) is really nothing more than Eq. (5) for
all combinations of two items, and Eq. (11) is a composite of many
equations like Eq. (9). We dispense for the moment with carrying the
story on to higher-order frequencies and turn to the main problem,
the solution of Eqs. (8) and (11). This, and the corresponding problem
of other models, has so far been the central concern of latent structure
investigations. The next four steps are devoted to it.

Step 3 : The conditions of reducibility. From the previous step
the following observation can be made. For m items, there are 3m
latent parameters, two coefficients for each of the m trace lines and one
moment for each of the m frequency levels which can be formed. But
we have 2m manifest data. Thus as the number of items increases we
will have many more equations than unknowns. This means that the
model imposes restrictions on the manifest frequencies. One could say
that each model determines the morphology of the dichotomous sys-
tem which it generates.

The third step in the latent structure analysis consists in studying
the dependencies which exist between the manifest frequencies accord-
ing to the special nature of the model.

These restrictions are called conditions of reducibility, for the
accounting equations can be solved only if these conditions are met;
or to put it differently, a dichotomous system to be reduced to a system
of homogeneous classes in accordance with a special latent structure
model has to satisfy these conditions. The difficulty at this step derives
from the fact that no general rules for the finding of these conditions
can be established. Each model has to be investigated separately, and
one usually ends up with a large number of conditions. It is much
more difficult to say what minimum set of conditions is sufficient to
determine all others. As a matter of fact, we will not tackle this prob-
lem at all here, though in the subsequent comments the answer will be
provided for the particular model under discussion.

One general lead can be given as to the nature of these condi-
tions of reducibility. They usually consist of quite complex combina-
tions of manifest data which on the latent side, however, are much

Latent Structure Analysis 513

simpler than the original accounting equations. Let us take as a first
example such a combination which can be formed within the matrix
Pof Eq. (6).

We select out of this matrix a special matrix by picking any two
rows and any two columns, but the selection is made so that the col-
umn numbers are all different from the row numbers. To be specific,
we might pick the first two rows and the third and fourth columns.
We then form a matrix of order 3 X 3 by adding a first row and a first
column which consist of marginals corresponding to the rows and
columns we have just selected. If \ve put 1 in the upper left corner, we
get the following form:

Pz P* >

B = pl pu pu (12)

We will say that the bordered matrix B has a vertical signature 1,2 and a
horizontal signature 3,4. It can be seen from the formation of such
matrices that we never have to worry about the missing diagonal
entries of the original matrix P. We can form as many such specific
matrices as there are combinations of four items in our reservoir of
items; in addition, from each such matrix as the one shown in Eq. (12),
we can always derive one essentially different matrix by exchanging
the identifying indices of one row and one column.

How does such a combination of manifest data look on the latent
side? The relation is very similar to Eq. (8); only we have now to deal
with two different matrices A, one corresponding to each of the two
signatures. They are defined in Eq. (13).

n i i \ ^H n

U ai a% ) ( 0

The accounting equation for the matrix B reads as follows:
1 0 1

B = A.'VMA.H =

Ml

,o

0 -i -i. W

From the expression on the right we know immediately that the de-
terminant of B must be zero. This follows from some theorems of ele-
mentary matrix theory which may, for our particular case, be stated
as follows: If a matrix W^is equal to the matrix product UV> then the
largest square submatrix of W with non-zero determinant cannot be
larger (i.e., cannot have more rows and columns) than the largest
square submatrix with non-zero determinant contained in either of

514 PAUL F. LAZARSFELD

the matrices U or V. Since none of the three matrices on the right of
Eq. (14) is large enough to contain a square submatrix with more than
two rows and columns, the matrix B on the left cannot possibly contain
a non-zero determinant of order greater than two. Thus one condition
of reducibility is clearly that any bordered 3X3 determinant which
can be formed from the full matrix of the second-order joint fre-
quencies vanishes. It can be shown in a similar way that for the
stratified matrix P& the same conditions hold.

In addition to these conditions, which hold for each level of
stratification, one can also deduce another set of conditions which
prevail between different levels of data. This can be shown by forming
determinants of order 2X2 taken from bordered matrices P and Pk
as defined in the previous section, Eqs. (6) and (10). We form sub-
matrices by picking just one row and one column bordered by mar-
ginals as before. Choosing for example row 1 and column 2, we get
now from P the form:

1 P* } - { 1 0 1 f 1 Afi ] f 1

Pi PU i [ 01° ai1 J (MI M, J lo

Similarly we get from Pk

f Pk P*k

( pik puk

^ 0

f 1 0 I f 1 Ml

fli1 J I Afi

i ,.o i 1
0

0 ak

Before going on we shall anticipate one point which will be dis-
cussed in more detail in the next step. It is not possible to develop a
complete metric in the latent space. The zero point of the latent con-
tinuum and its unit of measurement remain unidentifiable. We deal
with measurements similar to the temperature scale, where only ratios
of "distances53 have an intrinsic meaning. This means that without
loss of generality, we can fix the values of two moments; we might as
well simplify our computations by making MI = 0 and M2 = 1 . The
average position of the population is then at the origin of the coordi-
nate system and the standard deviation of the population distribution
becomes the unit of measurement. As a result the determinantal equa-
tions corresponding to Eqs. (15) and (16) acquire an especially simple
form. Using an obvious symbolism for their left sides, the right sides
become:

[12] = tfxW (15a)

[12;*] = <2iW[fe°)2 - (a*1)2 + aMM*] (16a)

Latent Structure Analysis 515

Forming the ratio of these two equations we find

[1 — j/Cj . 1,0 t ft 11 f

[12]

(17)

The essential feature of Eq. (17) lies in the fact that the latent form
on the right side contains only moments of the distribution function
and the traceline coefficients of the stratifying item. If on the left side
we had chosen any combination of items other than 1 and 2, the ratio
would still remain the same as long as item k remains the stratifier.
Once the frequencies on one level are fixed, there is little freedom left
in this model for frequencies on another level because there exists a
kind of proportionality between levels. It is easily shown that condi-
tions similar to Eq. (17) exist between any two different frequency
levels.

Step 4: Identifiability. Our next concern is whether the data of
our dichotomous system are sufficient to fix the values of the latent
parameters. From the previous step we know that it is obviously not
enough to count whether we have more equations than unknowns.
The conditions of reducibility have shown that in this (as in any other)
model, many of the manifest data are derived from others and there-
fore only a portion of the accounting equations can be independent.

The accounting equations, like (2) and (3), contain definite inte-
grals. These do not change in value under a large class of transforma-
tions of the x axis. However, if we want to maintain the linearity of the
trace lines, then only linear transformations affect neither the model
nor the data generated by it. But the fact that linear transformations
of the latent continuum do not affect the observable consequences of
the model implies that we will be able to identify the whole latent
structure only up to a linear transformation of the latent continuum.
For instance, if we fix the coefficients of one trace line arbitrarily, the
rest of the structure would be fixed.

Alternatively we can choose an arbitrary zero point and an arbi-
trary unit for the latent variable x. For a variety of reasons, it is in
general preferable to make the second choice. We set MI = 0, M% = 1 .
The implications of this in terms of traditional measurement theory
were explained in the previous step.

We have, so far, only shown that two latent parameters must be
arbitrary. We have not proved that all the others are identifiable.
This will be obvious as a result of the next step.

One should not confuse the problem of identifiability with the
question of whether enough items are available in a specific research
problem. The origin and the unit of measurement for the latent con-
tinuum cannot be found in this model irrespective of how many items

516 PAUL F. LAZARSFELD

are added. On the other hand, if we are interested in getting an addi-
tional number of higher moments we can always find them by adding
additional items to our reservoir of manifest data. In some latent
structure models it is not at all easy to find out how many items are
needed to identify a latent structure, or to identify it up to a given
point. In the present model it is very simple. As we shall see from
step 5, the coefficients of the trace lines can be found by using only
data of the first and second order. As we want to find more moments,
we have to move to ever higher frequency levels.

Step 5: Identification. The problem of solving the accounting
equations is somewhat similar to the task of finding the conditions of
reducibility. It again requires that we think creatively of combinations
of manifest data which, however complex on the manifest side, become
more simple in their latent form.

The traceline coefficients are very easily found by going back to
Eqs. (2) and (15). Let us remember that we have given the first two
moments arbitrary values: MI = 0, Af2 = 1. This puts Eq. (2) into
very simple form:

Pt = at0 (18)

In other words, the position of the trace line, its intercept with the
y axis, is immediately given by the marginal of the corresponding
item. To get the slope of the trace line we turn to Eq. (15). Forming
determinants, we get

1 0

.0

1 0
0 1

1 ay'

0 fly1

(19)

Whereas in Eq. (15) we used two specific items, 1 and 2, we now write
the result in general form for any two items, i andj. The symbol intro-
duced in the middle of Eq. (19) is a convenient representation for the
cross product, which occurred already in Eq. (15a).

Thus it turns out that the manifest cross product between two
items is the product of the slope coefficients of the two corresponding
trace lines.

How would we get this coefficient for the single item, say i? We
need two auxiliary items, b and c. Then we have the answer in the
equation:

Kib][ic]

(20)

It should be noted that Eq. (19) implies that it makes no difference
which two auxiliary items b and c we use. (Indeed, this is a condition
of reducibility.) We shall see in step 6 that when we deal with actual
data the situation is somewhat different.

Latent Structure Analysis

517

At this point we have in principle Identified the two traceiine
coefficients for all Items. How about die moments? The third moment
can obviously be obtained by making the same substitutions In
Eq. (16) as we made in Eq. (15). Without going Into details, we write
the result in Eq. (21).

\s [y>£] Pk , ^k1 /01x

A/3 = — - £-: + — (21)

Recall that 0&1 is already known from Eq. (20).

When it comes 10 the fourth moment, a new idea is Introduced.
It can be shown that none of the manifest forms we have mentioned so
far would ever help us to compute a fourth moment. We have to dis-
cover a new combination of manifest data for this purpose. It turns
out that so-called ascending matrices are the appropriate device. In
these the signatures contain elements of different orders, e.g.,

A =

Pi
pl2

J&134
^1234

(22)

When more than four items are under consideration, stratified ascend-
ing matrices also play a role. We can write a matrix equation which
contains the accounting equations for all the elements in the ascending
matrix A in the form

where

and

A = (A*)'M3X3A.

MQ Mi

MI MI

A* =

0 ail au1
. 0 0 a^2

and similarly for

(23)

(24)

(25)

The elements af are the traceiine coefficients of the Jcih power of x for
the ith item. The elements a^k are the convolutions which we have
encountered already in Eq. (3). We shall not go through the details of
demonstrating Eq. (23) ; the reader will have no difficulty convincing
himself of the truth of it by carrying out the necessary matrix
multiplication.

Again by taking determinants, this time on both sides of Eq. (23),
and then simplifying, we obtain for the fourth moment the formula

(26)

518 PAUL F. LAZAJR.SFELD

The fifth moment can be obtained by stratifying the ascending
matrix we have just introduced. The sixth moment requires an ascend-
ing matrix with one more row and one more column, and so on.

As we mentioned before, the higher moments we want, the more
items we need. In our example we will be satisfied with the fifth
moment.

So far we have dealt with purely algebraic problems. We assumed
that the manifest data were generated by the model under investiga-
tion. We asked ourselves then how we could, so to speak, rediscover
the parameters of the model if we were only presented with "perfect35
manifest data. In actual research practice, of course, these data are
at least subject to sampling variations. Besides, most models under
investigation cannot be expected to be more than a rough approxima-
tion of whatever the "true" latent structure might be. The next step
requires dealing with empirical data.

Step 6: Computation: The fitting procedure. In the last section
we discussed the identification problem — the problem of solving the
accounting equations when the datum fits the model exactly, when it is
of the form which would be generated by the assumed model when
the sample size approaches infinity. With empirical data the situation
is never so clear. In empirical work it is from data beclouded by
sampling variability that we have to find the latent parameters.
The computation for the first traceline parameter a? is no more com-
plicated than the formula (18) obtained in the identification process
indicates. They are simply equal to the manifest marginals, pi = a?\

a* = .34 a2° = .57 a3° = .62 «4° = -48 a5° = .38 a&° = .58

The second traceline parameters a^- are a little more troublesome.
FromEq. (19)

[y] = *<V

we know that

[be]

Thus a*1 may be computed from the cross products. However, there are

9 cross products but only m latent parameters a*1. (For the case
t z j

of our six-item example there are fifteen cross products.) The result
of this is that there are a number of different combinations of cross
products which should give the same latent parameter a*1. These
different combinations will not have the same value if empirical cross
products are used in our computations. At present there are no stand-

Latent Structure Analysis 519

ards available on the basis of which one might judge when two of
these estimates of latent parameters are effectively equal.

In some way we must average the different estimates for the
parameter a? that can be computed from our data. The easiest way
to include all the cross products in the computation of the a^- is to
consider the equations of the form

= [ib}[ic\

(27)

If we add the equations for all possible combinations b,c, holding i
fixed, we may then factor out the (a{1)2 on the left and get ail as the
square root of the ratio of the two sums:

(28)

Although this appears quite complicated, there happens to be a very
convenient computing device in terms of more symmetric operations
than appear in Eq. (28). We first write out the cross product matrix
as in Table 5.

TABLE 5. GROSS PRODUCT MATRIX

1

2

3

4

5

6

1

.041

.062

.069

.057

.029

2

.041

—

.080

.088

.077

.050

3

.062

.080

—

.107

.088

.054

4

.069

.088

.107

—

.103

.061

5

.057

.077

.088

.103

—

.058

6

.029

.050

.054

.061

.058

—

Then simply by summing the columns and squaring and similar
operations and substituting in the computation formula7

(29)

we get the values written in the second line of Table 6. In its first line
we have repeated the values we found for a£.

7 This formula, which is algebraically equivalent to Eq. (28), is Spearman's
famous single-factor formula.

520

PAUL F. LAZARSFELD

TABLE 6. THE TRACELINE COEFFICIENTS

Item number £

1

2

3

A

5 i

!

6

a<°

ail

.34
.185

.57
.254

.62
.309

.48
.348

.38 !
.300 i

.58
.181

We now have all the necessary data for drawing graphs of the
trace lines, as we have done in Fig. 6. We are not surprised that the
graphs of items 1 and 6 are almost parallel, since their slopes, a^ and
fle1, are very nearly the same; and similarly, the trace lines of items 3
and 5 are just about parallel. None of the trace lines in the set can

-1.0

-0.5 0 0.5

FIG. 6. The trace lines.

1.0

1.5

meaningfully extend beyond the point where any one of them be-
comes greater than 1 or falls below zero — certainly the distribution of
respondents must be zero beyond those points. Linear trace lines give
us some insight into the relationship between the different items and
between each item and the latent continuum. (This will be discussed
in the next section.) The data of Table 2 permit us to construct
composite trace lines, those for response patterns consisting of more
than a single item. The principle of local independence ensures that
the trace line of a response pattern is given by the product of the
trace functions for the individual responses which make up the re-
sponse pattern. For example, the trace function for the four-item
response pattern, consisting of positive responses to all the items
3, 4, 5, and 6, is

/3456 = /s(*)/4(*)/5(*)/6(#)

= (.626 + .3009*)(.481 + .3454*) (.406 + .3002*) (.577 + .1833*)

= .006*4 + .046*3 + .131*2 + .158* + .070

Latent Structure Analysis 521

When a response pattern Includes negative responses, we have to
replace the trace functions corresponding to the iiems answered
negatively by their difference from unity, that is, we replace f(x) by
1 —/(#)• For example,

In Fig. 7 there are drawn the composite trace lines for the four re-
sponse patterns 46, 46, 46, and 46. Individuals at the high job-satis-
faction end of the latent continuum are most likely to respond
positively to both items 4 and 6, whereas at the other end of the latent

^46 ^46 ^46

FIG. 7. Trace lines for four response patterns.

continuum, a negative response to both items is most likely to occur.
This is what one would expect in advance. The really new insights
come if we compare the trace lines /4g and /46. Their form could not
be guessed by looking at the content of the items. It turns out that
/46 is more like /46 and /46 more like /4g. A more detailed analysis
would show the reason: item 4 (as can be seen from the slopes in Fig.
5) has a sharper relationship to the latent continuum, is more indica-
tive of it; as a result it, rather than item 6, determines the place of
H — and — h between the consistent response patterns + + and

. The matter of ordering will be taken up in step 9, and then we

shall also explain the other features of Fig. 6 not yet discussed here.

The basic and the composite trace lines represent conditional
probabilities. They indicate how likely a person is to exhibit a given
response pattern if he is at a point x of the latent continuum. This
still leaves the question open, how many respondents are at each
point x. To answer this we need the density function <p(x) for the
whole population of respondents. What we can identify are the
moments of this distribution.

522

PAUL F. LAZARSFELD

For the computation we can use fitted cross products a^af- or
fitted second-order positive frequencies a£a^ + a^a^ rather than the
values obtained directly from the data. We thus eliminate some of
the random variability. We can, for example, substitute these fitted
values into the expression for M$ [Eq. (21)]. Only three items are in-
volved in any one computation of the third moment, so that „

20 different computations are possible. In practical work we average
either all of these computations or some sample of them. In our
example the average of the estimates of M% turns out to be —.011.

From the accounting equations for third-order parameters we can
now compute third-order fitted frequencies. Then in the computation
of the fourth moment we can again use fitted parameters, instead of
raw data, at all levels up to the third. On the other hand, there are

= 15 new pieces of empirical data, the 15 fourth-order joint

positive frequencies, each of which provides us with one estimate of
the fourth moment. Again averaging all these computations, we get a
value of 1.57 for M^ and finally a value of 5.47 for M^

From four moments one can get a rough idea how a distribution
looks as compared with the well-known normal distribution. M% is
slightly less than zero, which means that the distribution is somewhat
skewed to the right. M± is considerably less than 3, which means that
the curve is much flatter than the normal distribution. With higher
moments we can compute equivalent discrete classes. This is a pro-
cedure which is of importance for many latent structure models and
therefore deserves further special mention.

Suppose we want to approximate the distribution of people by
assuming that they are concentrated at three points %i with a relative

frequency of vt so that \ vi == 1- We can then define two moment

matrices

M =

MQ MI
MI Mi

M* =

MI

Ms

It can be seen easily that

where N =

M =

WNW M*

0
0

0
0

0

V

and

= WNXW

xi 0 0

0 X2 0

0 0 #3

Latent Structure Analysis 523

and W =

1 1

xl %2

If we form the determinantal equation

\M* - xAf\ = 0

its roots can be sho\vn to be the location points #,-. Computation shows
that practically all the cases cluster in two classes located at ATI = —1.03
and #2 = .96. In other words, the data indicate that people fall into
about two equal classes, those who are satisfied and those who are
not satisfied with their jobs; the frequencies in the two classes are
respectively vi = .482 and #2 = .517. A third class of completely in-
significant size is characterized by very extreme satisfaction.

Step 7: Evaluation of the fit. After we have found the latent
parameters we can ask: how well do the data agree with the fitted
model? Since the data will never fit the conditions of reducibility
exactly, we cannot expect that the "fitted frequencies'5 will be identical
with the data. On the other hand, we should expect the differences
to be small, and randomly distributed.

In most latent structure models, the method of solving the ac-
counting equations proceeds in a manner similar to that described
for the linear traceline model: the manifest frequencies are taken into
account level by level. In our example the first-order frequencies pi
did not require any manipulation; we simply accepted them as
estimates of the latent parameters <zA Then, by a certain averaging
process we found the second traceline parameters a£ from the cross
products. But now we can already compute fitted cross products and
fitted second-order frequencies from the traceline parameters; and
before going on to find further latent parameters involving higher-
order manifest frequencies, we can make some evaluation of how well
the data up to the second-order frequencies are in agreement with
the model. If we decide that the fit at this level is good enough, we go
on to find other latent parameters, using wherever possible fitted
frequencies instead of manifest data. For some models, including the
linear traceline model, it is possible to make successive evaluations of
the fit after each higher level of data has been utilized in the com-
putation.

Since from Eq. (19)

[ij] = a^af

524

PAUL F. LAZARSFELD

we can compute a fitted cross product matrix from the parameters
In our example we obtained the following:

TABLE 7. FITTED GROSS PRODUCT MATRIX a^aj1

1

2

3 4

5

6

\

.047

.057 .064 .056 .033

2 i .047 —

.078

.089 i .077 .046

3 .057 1 .078

—• .108 .093 ! .056

4

.064 ! .089

.108

—

.105

.063

5

.056 ! .077

.093

.105

—

.054

6

.033

.046

.056 .063

.054

—

Is the fitted cross product matrix ccclose enough55 to the matrix
of cross products obtained from the data? The simplest way to com-
pare the two is to subtract one from the other, and to consider the
size of the residuals:

TABLE 8. MATRIX OF RESIDUALS [if] — a^af-

1

2

3

4

5

6

1

—

-.006

.005

.005

.002

-.004

2

-.006

—

.002

-.001

.001

.005

3

.005

.002

—

-.001

-.005

-.002

4

.005

-.001

-.001

—

-.002

-.002

5

.002

.001

-.005

-.002

—

.003

6

-.004

.005

-.002

-.002

.003

—

The relatively largest residual is —.006; the corresponding em-
pirical cross product is .041, which makes the residual just about 15
per cent. It can be shown that these residuals have a concrete meaning,
which can be understood best by reference to Fig. 1. There we saw
that if we use a raw score to simulate a latent continuum, the associ-
ations within each of the partial fourfold tables do not vanish. If
our data fitted the linear traceline model perfectly, the principle of
local independence would assure us that no residual associations of
this kind remain. If they do, they indicate the average cross product
remaining between two items after the latent continuum "has been
taken out.53

Once the model parameters are found, we can compute the fit
of the model on any level. Table 9 gives the actual and the fitted
positive joint frequencies on the fifth level. As can be seen, the diver-
gencies are very small; but no theory of error yet exists to permit a
rigorous test.

Latent Structure Analysis 525

TABLE 9. ACTUAL AND FITTED JOINT FREQUENCIES OF ITEM QUINTUPLETS

Combination of items Actual joint frequencies Joint frequencies required by

J ^ model

12345

.127

.129

12346

.132

.134

12356

.113

.115

12456

.112

.103

13456

.119

.114

23456

.174

.178

Step 8: The recruitment pattern. The trace lines tell for each
point on the latent continuum the probability with which any re-
sponse pattern will occur.

Now we want to raise the reverse problem. Given that a person
exhibits this response pattern, where in the latent space is he located?
The answer is somewhat surprising. He can come from anywhere
in the latent space. But the probabilities are great that he will come
from certain sections and slight that he will come from others. Each
response pattern has its recruitment pattern, a distribution of "inverse
probabilities55; it indicates for every point x the probability that a
respondent with the given response pattern comes from this place
in the latent space. It is important to understand the difference be-
tween traceline probabilities and recruitment probabilities. Perhaps
the simplest way to illustrate the difference is to consider a cross
classification of the population of respondents by response pattern
and location in the latent continuum — the distribution of people
being, for the sake of illustration, considered as a set of discrete
classes. In the table below let ngs be the number of individuals giving
the response pattern g who are in latent class s, ng being the total
number of individuals giving the response pattern g, and ns being the
total number of individuals in class s, and n being the total number of
individuals in the population. The traceline probabilities are the

ttl1

Manifest
response
ng pattern
frequency

j2 . . . ns ...

Latent class frequency

526 PAUL F. LAZARSFELD

probabilities of response pattern g at given points of the latent con-
tinuum, in our table the probabilities in a particular latent class:

*•'-*

On the other hand, the recruitment probabilities are the probabilities
of a latent class s, given the response pattern g:

71s

Obviously rgs = — * pf

ng

The probability of a response pattern g coming from a class s is
directly proportional to the size of class s and the latent probability
of g at S) and inversely to the frequency of the response pattern. If we
have a continuous distribution, we can define

(30)

Po

where pg is the proportion of all respondents giving response pattern
g, and fg(x) is its composite trace line. Equation (30) tells us to what
extent respondents of type g are recruited from each point x of the
latent continuum. But we know only the moments of <p(x). Therefore,
we can only compute the moments of ^fg(x). As an example, let us
consider the two-item pattern of positive responses to both items 4
and 6. The mean position may be computed as follows:

/ x*u(x) dx~ — I */4e(*)0(*) dx
J P^J

= -~ f x(ABl + .345*)(.576 + .183*)0(*) dx

= .B17Mi + .847M2 + .186M3 = ju46

and substituting the numerical values which we have found for the
moments

M46 = .835

Similarly we would find that

dx = .441

P&
and jaj6 = -.433^33 = --859

Latent Structure Analysis 527

The values of these expected recruitment positions are Indicated
in Fig. 7. It Is important to notice that we need \ 'm -f 1) moments and
therefore (m + 1) items, if we want to compute this position for a
response pattern based on m items. It is also possible to develop a
measure for the discriminating power of a response pattern g in the
form of

fx*tyff(x) dx

but Its discussion would lead us too far afield.

Step 9: Classification and scores. When we have found out as
much as we can about the distribution of respondents over the latent
continuum for each pattern of responses, w?e may then ask how we
might assign the individuals who have responded in a particular way
to a point in the latent continuum. We might ask about the most
typical, the most likely, or the "average33 position of the respondents
of a particular type; or we might be satisfied with an ordering of the
response patterns on the basis of some such criterion. In the preceding
section we talked about the mean values of x for given response
patterns, and they are often convenient as indicators of "typical
position" for respondents, or as Indicators of the rank of the response
pattern along the latent continuum. An alternative is to ask, what Is
the most probable position for an individual who has responded in a
given way. This question clearly can be answered only in those cases
where we know not just a few moments but the whole distribution
of recruitment probabilities given by Eq. (30).

Mean values, or modal values, such as we have here discussed may
be looked upon as scores to be assigned to the respondents on the basis
of their response patterns. But whenever we construct scores for
respondents or their responses, there is a corollary problem of scoring
the Items — how much does each contribute to the score of respondents,
how much does each item help to discriminate between individuals
located at different points of the latent space? We remark only that
in the linear traceline model it is clear that the steeper the trace line
the better it can discriminate between two extremes of the range of
the latent variable. The steepness of a linear trace line is indicated
by its slope, a£. For other models, similar indices may be constructed,
though they usually do not fall out of the latent structure in such a
simple fashion. [See 14, p. 377, for more detailed discussion.]

Summary. We have now gone through the nine steps of latent
structure investigation for the linear traceline model. The outline is
the same for all latent structure models and is presented schematically
in the diagram following.

528

PAUL F. LAZARSFELD

STEP I

General notions about
data of this type:

(a) substantive

(b) formal
Specification of the
model

-»| 2nd order"] —

Parameters depending
only on first order
data

Parameters depending
only on 1st or 2nd
order data

Parameters depending
on data of all orders

Fitted first
order data

Fitted first
and second
order data

Rtted data of
all orders

If unsatisfactory
return to STEP I

If acceptable

FIG. 8. Summary of latent structure steps.

IV. THE PROMISES AND LIMITATIONS OF LATENT STRUCTURE
ANALYSIS

Although the linear model is of unusual mathematical simplicity,
the steps which were described in the preceding section are typical
for all models; so are the findings. A complete solution consists of the
following elements: (1) the coefficients of the trace lines, (2) informa-
tion on the distribution of people over the latent space, (3) indications
as to how well the assumed model fits the empirical data, (4) pro-
cedure to score response patterns if such scores are desirable.

The question now arises: what scientific contributions can latent
structure analysis (LSA) make? Two aspects have to be distinguished.
One is the possible contributions to the logic of empirical research.
This is best discussed by means of a comparison with procedures which
have a similar intent; we shall presently turn to such a comparison of

Latent Structure Analysis 529

LSA with factor analysis and formal test theory. The other aspect
Is the practical usefulness of the numerical results obtained from a
specific model.

The meaning of trace lines. At this point it is necessary to warn
against overrating models of this kind. They do not represent a theory
In a strict sense, but a set of organizing principles. A typical investiga-
tion of an attitude or trait begins with the assumption that certain
Indicators will be useful to classify people for a given purpose. LSA
can only clarify intrinsically the meaning of these indicators. It can-
not tell whether the general purpose of the investigation has been
reached. The nature of such an Intrinsic analysis is best approached
through a concrete example taken from a somewhat more compli-
cated model.

In a study of academic freedom, a so-called apprehension test
was developed. Social science teachers were asked a number of
questions, of which the following four are characteristic examples:

1. Have you worried about the possibility that some students might in-
advertently pass on a warped version of what you have said and lead to false
ideas about your political views?

2. Do you ever find yourself wondering if because of your politics or
something political you said or did that you might be a subject of gossip in the
community?

3. If you are considering a move to another college, have you wondered if
that college would inquire at your present college about your political views?

4. Have you toned down anything you have written lately because you
were worried that it might cause too much controversy?

The model applied to these and a number of similar items was a
so-called latent content model. The trace lines of this model are a
special case of the following equation:

If in this equation the exponent d approaches infinity, then the model
formalizes the well-known social distance scale developed by Bogardus.
If we specify further that a = 0 and b = 1, we have what is called a
perfect Guttman scale [26].

On the other hand, if we set c = 0, we have a trace line with three
parameters which is able to summarize a great deal of useful infor-
mation. 8 Figure 8 shows the trace lines for the four items in the appre-
hension index just quoted.

The interpretation of these trace lines is almost obvious. The first
three items happen to have almost the same marginal response fre-

8 For the details of this model, see [25],

530

PAUL F. LAZARSFELD

quency — about 40 per cent. But the relation to the latent continuum
varies. The gossip item (2) is represented by a practically straight
line. As teachers become more apprehensive, they are more likely
to be concerned about the repercussions of their political views in the
community; and this probability increases quite proportionately to
the increase in apprehension. Concern about one's future job (3) be-
haves differently. Its probability rises much more quickly; already at
a low degree of apprehension teachers are likely to worry that their
chances to move to another college would be jeopardized by some
opinion they have expressed at their previous job. On the other hand.

i.o

.2 0.8

1
o

Q.

§ 0.6

0.4 -

» 0.2

Students {/)

Toning down (4}

0.2 0.4 0.6 0.8 1.0
Latent continuum: apprehension

FIG. 9. The trace lines for four items of an apprehension index.

even at a high degree of apprehension this worry is not as general as
the concern with gossip. After all, gossip is an ever-present danger,
although many people do not think far into the future or do not ex-
pect that they will ever have to move to another college.

The item regarding student misrepresentation (1) is different in
two respects. First, we notice that the onset of its trace line is higher.
Even people who are not apprehensive at all reckon with students3
misrepresentation as part of the necessary hazards of their occupation;
for quite a while as apprehension goes up, the probability of this con-
cern does not increase very much. But at a very high level of appre-
hension it suddenly shoots up and becomes rather dominant. The
fourth item deals with the toning down of one's own writing. Here,
too, the probability that a teacher tells about such a precautionary

Latent Structure Analysis 531

move appears only on the right side of the graph and then rises very
rapidly. But compared with the others, the manifest frequency of this
item is much lower. It will be remembered that the marginal fre-
quency is essentially an integral over the trace line and therefore in
the graph represented by the area below it.9

By now the reader should realize clearly that this graph is not the
result of a conventional item analysis. The underlying continuum of
apprehension is not represented by a raw score of any kind. The
parameters of the trace lines are derived from the higher-order joint
frequencies, the manifest data which describe the interrelation be-
tween the items. The trace lines, so to say, define the meaning of the
underlying dimension. But, at the same time, they clarify the meaning
of the various items in relation to each other. Each of the three coef-
ficients a, by and d makes a different contribution to the shape of the
trace line y = a + bxd. b indicates something like the expressive
value of the indicator. The larger b is, the greater is the difference in
probability between the left and the right side of the graph; this
means that in respect to this indicator, apprehensive and nonappre-
hensive teachers are especially different. The coefficient a tells to wThat
extent an affirmative answer is common to all teachers irrespective
of their own apprehension. The curvature of the trace line is approxi-
mately indicated by d; it could be called the severity of an item:
whether an affirmative answer is given easily or whether it needs a
great deal of apprehension to reach it. Here we have an obvious
parallel to the notion of difficulty in knowledge tests.10

The clarification of meaning is then one of the major results of
a latent structure analysis, and this turns out to be a rather complex
procedure. The underlying continuum and the psychological mean-
ing of specific questionnaire items define each other. The same item
combined with the different set of others could have a different trace
line, and therefore contribute different meanings to the whole struc-
ture. This, however, is exactly what we should expect in material
which has a strong projective element. After all, whether a teacher is
worried about misrepresentation by students can be the indicator of
a politically endangered professional situation. But it could also be
the expression of a general trait of anxiety. With questions pertaining

9 The latent content model includes the assumption of a uniform distribution of the
population of respondents over the latent continuum. See discussion below.

10 LSA refines the traditional notion of difficulty. Two items in a test can have the
same manifest frequency of correct answers and, therefore, the same area under the
trace line. They, however, could differ in shape, like items 1 and 3 in our graph. Item 1
would be more difficult in terms of the specific ability on the test but easier as far as
common knowledge goes.

532 PAUL F. LAZARSFELD

to professional situations, an affirmative answer might have a different
significance than if the item is combined with other questions per-
taining to nonprofessional concerns. Consequently, the value of a
latent structure analysis is considerably greater with somewhat
ambiguous matter than with questions where the content is clear-cut
— for instance, the items in an arithmetic test.11

Before we leave the example, one other aspect of this traceline
graph should be mentioned. At the left and the right end, some of the
probabilities go below 0 and above 1, respectively. This, of course,
is a mathematical absurdity. The extent to which it happens indi-
cates that the model is not quite appropriate to the data, either be-
cause it is too simple or because of sampling errors. A sampling theory
has been developed for some models, but by no means as yet for the
system as a whole [3, 19].

Distribution in the latent space. So much for the practical import
of the trace lines. The information provided by the distribution curve is
more obvious, but it is worth while to distinguish four types of situa-
tions. In the model discussed in section III, no assumption was made
about the latent distribution. We were able to compute some of its
moments; and it became clear that the more items are available in
the manifest data, the more moments could we compute. In a more
generalized form the trace lines in this model would be polynomials
of any degree, and the latent space could be multidimensional. The
algebra of this model has been solved completely and represents the
most advanced point of LSA at the moment [20, 22].

In a second group of models one assumes a rectangular distribution
of the population, say between the limits of 0 and 1. This is really
tantamount to abandoning any effort to find a metric in the latent
space and looking for only an ordinal ordering of response patterns.
One might call this group scale models. The classical example is the
Guttman scale and a variety of possible generalizations corresponding
to what Guttman calls quasi scales. The most manageable model of
this kind is the so-called latent distance scale [20, 26],

A third type of assumption is very suggestive but leads to extraor-
dinary mathematical difficulties. Certain algebraic forms for trace
lines are chosen because they approximate what we think is the be-
havior of people who differ in their position on the latent continuum.
In the same spirit, we really are interested only in some general
information on the distribution of these people: whether it is right
or left skewed, steep or flat, etc. Such knowledge could be obtained
by giving the distribution function a predetermined algebraic form

11 This is the reason why the model leading to Fig. 5 is referred to as a latent con-
tent model.

Latent Structure Analysis 533

with just a few parameters, to be computed from manifest data. For
instance, the form

<p(x) = kxa\\ — x)b

has a great deal of flexibility. The trouble however is, as we know from
the previous section, that such a function gets multiplied by the
formulas of the trace lines. In order to be able to evaluate the ensuing
integrals, the trace lines and the distribution function have to have
somewhat comparable form. Quite a number of combinations have
been tried, but none turned out very successfully. So far this is the
point where the least progress has been made.

From a practical point of view the most useful case thus far studied
is the discrete class model. It is best understood as a kind of latent
typology. The assumption is that people are divided into homogeneous
classes without implying any special ordering in advance. W. Gibson,
for instance, has taken data on preferences of radio listeners for thirteen
types of evening programs [23]. He has shown that the manifest
joint response frequencies can be reduced very successfully by assuming
six types of listeners. To simplify the presentation we reproduce the
latent probabilities for four of the latent classes pertaining to six of
the programs. The data of Table 10 suggested to Gibson a rather
convincing interpretation.

TABLE 10. LATENT PROBABILITIES FOR PROGRAM PREFERENCES IN FOUR LATENT

CLASSES AND MARGINAL MANIFEST FREQUENCIES (pi) FOR A SAMPLE

OF 2,200 RADIO LISTENERS

Cl

ass

Manifest

A

D

E

F

marginal

Comedy

37

58

85

96

62

Mystery

.34

29

83

22

.46

Serni-rlassical rmisir. .

08

92

41

09

37

Classical music

.10

.89

.25

.27

.34

Religion

.04

.32

55

61

.21

News

.40

.84

.80

.96

.77

Proportion of people in class

27

21

10

11

1 00

The most outstanding single characteristic of Class A is that none of its
latent marginals is greater than .50. This class must, therefore, consist princi-
pally of people who do not care much for listening to the radio in the evening.

Class D is also not difficult to identify, for a very high proportion of its
members, in contrast to those of other classes, like to listen to semi-classical
and classical music. (They are also fairly high on talks on public issues.)

534 PAUL F. LAZARSFELD

These are undoubtedly the sophisticates or "high-brows." Characteristically,
they have also little interest in mystery programs.

In contrast with Class A, Class E is characterized primarily by consistently
high latent marginals, none of which is lower than .25. (Substantial propor-
tions of this class even like to listen to serials and hillbilly music, which are
quite unpopular with all other classes.) It is interesting to note further that
the two more serious music programs liked least by this group are the same
ones that are liked most (except for news) by the sophisticates. All of these
characteristics suggest that this might be the "low-brow" group.

Finally we come to Class F, which is distinguished by the highest latent
marginal for religious programs (the lowest for popular music), quite low
ones for comedy and mystery, and no extremely strong likes other than for
news. This combination of a religious component, a lack of interest in what
might be regarded as a younger type of program, and a somewhat subdued
enjoyment in radio in general, except for news programs, points toward one
large group of radio listeners — that of older and/or small town people.

To test his interpretation, Gibson selected respondents who had a
high recruitment probability of coining from one of these classes and
studied their demographic characteristics and other information avail-
able about them. He found indeed a clear educational difference be-
tween Classes D and E, many more older people in Class F, and so on.
It should be noticed, incidentally, that, correlative to this interpreta-
tion of the classes, Table 10 also throws light on the latent appeals of
the programs themselves; the rather universally high preference for
news programs is here the best example.

The discrete unlocated class model has the advantage that only up
to third-order joint frequencies are needed to compute all latent
parameters. The procedure requires essentially the solution of de-
terminantal equations of the type used in section III, step 6. A corre-
sponding shortcoming consists in the fact that (in contrast, e.g., to the
linear traceline model) no ordering of the classes can be derived from
this model. Sometimes, however, an order is strongly suggested by the
content of the items together with an inspection of the latent proba-
bilities. The following example is revealing.

The reader is certainly acquainted with the type of attitude
measurement developed by Thurstone and Chave. The procedure
consists in submitting to judges a series of statements regarding, for
instance, Negroes. The judges classify these statements in an order
which ranges from strongly favorable to strongly unfavorable. From
this range obtained by ^ht judges a scale value for each item is derived.
Then a group of respondents checks the statements corresponding most
nearly to their own attitudes. The final attitude score of each re-
spondent is the average scale value of all the statements he checks.

It is obviously possible to treat the responses of such a test like

Latent Structure Analysis

535

any other set of response patterns and to submit them to latent struc-
ture analysis. Eight such items were selected and a latent three-
class structure fitted the manifest data well [8]. For our present purpose
we shall select five of these items. In Table 1 1 we report the wording
of the statements and the scale value which they got from the judging
procedure.

TABLE II
Statement Scale values

1 . The idea of contact with the black or dark skin of the Negro excites hor-
ror and disgust in me. 10.2

2. I am not interested in the Negro or in his relations to the white man be-
cause I think that in the end economic factors will decide his fate. 5 . 6

3. Sometimes I feel that the Negro is not getting a square deal, but at other
times I fee! he has a lot to be thankful for. 5.4

4. Until the Negro has been given more time and opportunity of showing
what he is capable of doing, I think that it is foolish 10 try to judge him. 5 . 1

5. I think that the Negro ought to be given every opportunity of education

and development — just like the white man. 1 . 1

It will be seen that two of these statements belong to the anti-
Negro and pro-Negro extremes, respectively. The other three, ac-
cording to the judges, have about the same middle position.

From latent structure analysis we can infer whether the respondent
actually did interpret the statement in the way the scale value indi-
cates. Table 12 gives the latent marginals for these five statements in
three latent classes: Class I is the most anti-Negro, and Glass III is
the most pro-Negro.

TABLE 12. LATENT STRUCTURE OF AN ANTI-NEGRO TEST

Reaction to item

I (anti)

II

III (pro)

Judges' scale values

1 Horror . . .

486

236

072

10 2

2 Economics

124

199

021

5 6

3. Sometimes

.385

.963

.276

5.4

4. Don't judge

.084

.489

.422

5.1

5. Give opportunity. . . .

.068

.667

.916

1.1

We see that the latent class structure corroborates the scale value
on the two extreme items. The idea that contact with a dark skin
excites horror gets its largest endorsement in Class I, and hardly
appears in Glass III. The willingness to give the Negro all possible
opportunity is inversely endorsed by Glass III and rejected by Class I.
On both items Glass II takes an intermediate position.

But how about the three middle items? In regard to item 3, Class
II has an almost complete propensity to endorse it. The probability

536

PAUL F. LAZARSFELD

of endorsement is small In Classes I and III. Looking at the wording of
item 3, we are not surprised. It really states in so many words that
the respondent cannot make up his mind on the Negro question.

Item 2 and item 4 however show a different structure from each
other and from item 3. In both of these items the latent marginals in
the middle class are not much higher than in one of the others. Item
4 has a high marginal in Class III while item 1 is relatively high in
Class I. What lead do these figures give for a better understanding of
the content of items 4 and 2? Item 4 states that it is too early to judge
the capabilities of Negroes. This could mean one of three things: One
should not judge him too hastily as capable, not too hastily as in-
capable, or just not judge him at all. The latent structure suggests
that the respondents in this group endorse this question mainly when
they want to prevent an unfavorable judgment. Item 4, in spite of
the medium scale value, really belongs on the more favorable side of
the presumed scale. This incidentally, will be understood if we add
that this test was given in the South where the prevailing trend is,
of course, unfavorable judgment.

Item 2 also contains two elements. One expresses a lack of interest
in the whole question; the other expresses faith that economic factors
will settle the issue. One might have expected that this statement
would have a greater appeal to liberal respondents who would be
likely to stress the importance of economic factors. This seems not the
case, however. It is the more discriminatory group which is likely to
endorse item 2. A possible interpretation is this: the statement starts
with the phrase "I am not interested in the Negro . . . "; for
many respondents the lack of interest in the issue might have been the
leading element in their interpretation. And in the South an unwill-
ingness to discuss the Negro question would, of course, be an endorse-
ment of the present bad situation.

Three questions then, which seemed to a group of judges to repre-
sent a very similar position, were not experienced in this way by the
respondents. Item 3, which explicitly expresses doubt, was a real
middle item. But item 2, which really was compounded of two state-
ments, seems to have given to the respondents more emphasis to one
of its elements, while the judges considered it balanced. Item 4 was
answered in a context which made it function less as a middle item
than its grammatical form led the judges to expect.

It deserves notice that Thurstone was aware that something like
LSA had to be tried. After presenting his judging procedure he wrote
[27]:

We shall mention here in passing the possibility of determining the scale-
values of the statements without the rather laborious sorting process. It may

Latent Structure Analysis 537

be possible to scale the statements directly from the voting records of a large
group of subjects provided that a considerable range of attitudes is represented
in the group of subjects used for this purpose. The principle involved is that if
two statements are close together on the scale, then the people who vote for
one of them should be quite likely to vote for the other one also. If the state-
ments are very different, spaced far apart on the scale, then those who vote
for one of the statements should not be very likely to vote for the other one
also. It might be possible to reverse this reasoning. We might then be able to
infer the scale separation between two statements in terms of the number of
subjects who indorse both statements, wi.2, the number who indorse the first,
72 1, and the number who indorse the second, 7i2.

The discrete unlocated class model serves well also if one wants to
analyze the type of work sponsored by Lloyd Warner. He, as is well
known, assumed that six classes are necessary to give an approximate
idea of American social structure. He uses itemized material like
people's properties, reading habits, and organizational membership
to allocate them into proper classes [30]. By using the joint higher-
order frequencies of these items, one can decide how many classes
reproduce the empirical data. For the material he has provided so far,
three classes seem to be sufficient [24].

So far all the examples have dealt with attitudes and therefore
necessarily pertained to individuals. It is easy to see how we would
extend LSA to collectivities. Suppose we were interested in the
cccohesiveness55 of a number of small groups. We might ask their
members a series of questions: Are most of their friends inside
the group? Do they like the group activities? What other group
would they rather belong to? By proper manipulation of data each
group could be classified in a manifest dichotomous property space
according to whether it was, say above or below the average on these
criteria. The ensuing response patterns could then be analyzed by any
latent structure model; the only difference being that the statistical
unit is a group and not a person. The same would be true if we took
cities and wanted to classify them by "goodness of life," taking as indi-
cators, e.g., the number of playgrounds, the number of libraries, the
juvenile delinquency rate, etc. Once these indicators are somehow
dichotomized the analytical machinery is exactly the same, whether
applied to people or cities.

It is more difficult to predict whether LSA will be useful if applied
to conceptually more complex intervening variables as they appear,
e.g., in learning theory. No effort in this direction has yet been made.
There is, however, considerable material available resulting from the
application of LSA to repeated observations. Processes going on in
time can be clarified this way. A simple example can be taken from a

538 PAUL F. LAZARSFELD

readership study. Three times, at intervals of a month, 986 people
were asked whether they had read the last issue of a weekly magazine
W. The manifest data were as follows:

TABLE 13. READERSHIP OF MAGAZINE W

Query period

Readership

Time 1

Reader
Reader Nonreader

30 11
14 60

Nonreader
Reader Nonreader

17 64
56 734

Time 2 .

Time 3:
Reader

Nonreader

It can be seen that 30 people had read each of the three issues
under study and 734 read neither; the others read some. From such
data the latent parameters of the following model can be computed.
It is assumed that people fall into two groups: readers and nonreaders
of magazine W. The (latent) readers have a probability to read a sin-
gle issue p1, which will be large but not quite unity. The nonreaders
will have a small (latent) probability p2; this means that even non-
readers look at an issue occasionally. Under this assumption it is possi-
ble to derive that 6 per cent of the sample are latent readers and 94
per cent are not. The former have a probability of/?1 = .75 to read a
single issue, for the latter/?2 = .08. This means that nonreaders behave
more predictably than readers, which makes intuitive sense. L. Wig-
gins has developed quite complex dynamic models and has applied
them to repeated observations on voters, consumers, etc. [3 1].12

Comparison with factor analysis. One obvious difference be-
tween LSA and factor analysis lies in the manifest data with which
they deal. The raw material of factor analysis is the quantitative score
which presumes already that a number of qualitative items have been
combined into a test. The latent structure analyst starts with the test
items themselves. Now it is true that factor analysis has been used for
qualitative items by applying various coefficients to measure associ-
ations between fourfold tables. But it is well known that a factor struc-
ture can come out differently, according to whether one uses tetra-
choric correlations, point coefficients, or any other such device. Of
this difficulty, LSA is free because only independence enters the pic-
ture, and this can be defined without using any measure of association.
The principle of local independence reformulates the whole problem
so that we never need to use any measure of association between the
manifest data. It is important to realize that the cross products and

12 The main results are to be reported in a forthcoming book on panel analysis.

Latent Structure Analysis 539

similar combinations of manifest data are only transient algebraic
devices to achieve the final result and in no way appear as substitutes
for correlation coefficients.

Green has shown that factor analysis can be reformulated so that
it turns out to be a special case of a general latent structure scheme [9].
In this version the factors form a latent space. At each point of the
latent space people have latent test scores and the Pearson correlation
between them vanishes. The actually observed correlations between
manifest test scores are derived by the mixing of latently homogeneous
groups. There even exists a complete parallel to trace lines: they are
formed by the average test scores at each point of the latent space.13
This reformulation of factor analysis would permit its extension to
nonlinear material. For nonlinearity is one of the main features of
LSA and this point deserves some further discussion.

The origin of factor analysis was Spearman's one-factor theory.
He assumed the scores of all tests to be linear functions of this one fac-
tor. The ensuing structure is very similar to the model we used all
through section III. When Thurstone extended Spearman's theory,
he added more factors, but the test scores remained linear functions of
them. It is however obvious that a different kind of extension of the
original Spearman theory would be possible. We could retain the
restriction to one factor, but the test scores could be nonlinear func-
tions of it. An example is the model exemplified by Fig. 8. Finally, one
can combine several latent dimensions with nonlinear trace lines. In
LSA it is possible to develop separate criteria, for the number of
dimensions and for the degree of nonlinearity of trace lines. The
ascending matrices mentioned in step 5, section III, are the crucial
device for this distinction, but it is not possible to extend this paper to
such detail [22].

Even a cursory reading of section III will have shown another way
in which LSA, while using the basic logic of factor analysis, extends
its range. In factor analysis only zero-order correlations among mani-
fest data enter the picture; they correspond to our second order joint
frequencies or cross products. Because of certain basic assumptions,
partial correlations do not add new information for the factor analyst;
they are an arithmetical derivation of zero-order correlations. To put
it differently, factor analysis does not make use of higher-order
covariances between test scores. Remembering the definition of a
correlation coefficient r12y one realizes that a form rm does not appear
in factor analysis; its definition would be

13 W. Gibson has extended this idea to a large number of empirical examples in a
series of yet unpublished papers.

540 PAUL F. LAZARSFELD

where the letters stand for standardized score for three tests. In LSA,
however, such forms exist in terms of the higher-order frequencies.
The use of manifest higher-order frequencies^, p^^ and so on, makes
for the much greater flexibility of the latent structure model. In
step 5 of section III we have seen how the higher-order frequencies
permit the computation of the latent moments. The coefficients of the
linear trace lines, however, could be computed by going up only to
second-order frequencies. In models with nonlinear trace lines the
higher-order frequencies are also needed to compute the trace line
coefficients. In a rather crude generalization one can say that the num-
ber of coefficients in the trace lines determines the level of manifest
joint frequencies which enter the accounting equations.

Comparison with test theory. The full use of manifest data also
characterizes the difference between LSA and test theory. A test score
is a combination of a number of response patterns. All the people who
give an affirmative answer to say three out of ten items have the same
score. In LSA a distinction is made between the proportion of people
who give a positive response to any specific combination of three items
out of ten. In a test of n items, test theory uses only n + 1 manifest
frequencies while actually 2n are available. The notion of a response-
pattern score was explained in section III, step 9: it is the expected
position on the latent continuum of a person who gives a certain re-
sponse pattern. The test score, on the other hand, is the number of
items to which this person gives an affirmative reply. Interestingly
enough, a mathematical relation can be established between the
latent continuum and the test score. For a test score is a manifest
datum which has its own trace line. To show this important fact we
shall assume a test of just two items and use the data of Fig. 6 pertain-
ing to items 4 and 6.

At a given point of the latent continuum the probability of two,
one, or no affirmative answer is by simple probability considerations,
respectively:

•?<>(*) =

The values of these functions can correctly be called the latent scores.
Integrals of the form (l/pg)Jsg(x)x(p(x) dx give their expected values.
This is nothing else than the scoring procedure developed in steps 8
and 9, section III, extended to groups of response patterns which have
the same number of affirmative answers. To each test score then corre-
sponds a position on the latent continuum or, as we might call it, a
latent score. In our special example, these latent scores turn out to be

Latent Structure Analysis 541

the values /z4e and /i5g of Fig. 8 for the scores 2 and 0, respectively;
corresponding value for score 1 Is f^ieMie + p4&J>tf)/(pz& + P&)- Thus>
there correspond to the three test scores the following points on the
latent continuum:

^2 = .835 si = .008 s0 = -.859

They are not equidistant and in general the test score gives a distorted
picture of the latent continuum. From a practical point of view, the
distortion Is often unimportant. But the finding Is a good example of
the way a more general formallzation of a problem throws new light
on well-established procedures [17].14

One more relation between test theory and LSA deserves mention.
The difference between an item curve and a trace line was discussed
in section II. The x axis for an item curve (see Fig. 5) Is the raw score,
a manifest piece of information; for a trace line the x axis is a con-
struct, the latent continuum (see Fig. 8).

The consequences of this difference can be seen easily in the case
of a one-dimensional test. Within the framework of test theory we have
to assume in advance that the test items are indicators of a one-dimen-
sional continuum. This continuum is roughly approximated by the
raw score, and the shape of the empirical item curve is then used to
select items. In LSA, in the course of computing trace lines, we obtain
a simultaneous test of unidimensionality. It so happens that in the
linear traceline model this test is algebraically Identical with the way
one would test a Spearman one-factor structure. This., however, is by
no means true for all one-dimensional models. In the case of the latent
content model, for example, one-dimensionality requires the matrix of
the reciprocals l/[ij] of the cross products have rank two.15 Such tests
are, of course, only special cases of what in section III, step 3, were
discussed as conditions of reducibility.

The reader should keep in mind the difference between this com-
parison with test theory and the earlier comparison with factor
analysis. In comparing the latter with LSA, we mainly discussed
formal analogies and differences. Substantively the two procedures
deal with different material, except in the unfortunate case when fac-
tor analysis is applied to qualitative data. But fundamental test theory

14 The relation between latent continuum and test score is mathematically simple
in idea but cumbersome in expression. Frederic Lord has investigated it for a specific
model. His discussion and his concrete examples show many more implications of the
problem than we could touch upon here [17].

15 In [22], it is shown that this condition makes the model an interesting counter-
part of a two-factor Thurstone model. The "rotations" come about on hyperbolas in-
stead of circles and are determined by third-order frequencies.

542 PAUL F. LAZARSFELD

does not deal so much with relations between tests, but with the role
of the individual items within the same test; to this extent, it sub-
stantively overlaps with LSA and more work should be done in de-
riving notions like reliability or attenuation from the general proper-
ties of trace lines, latent probabilities, and latent spaces.

Be it repeated, however, that the application to one-dimensional
tests is not the only, and probably not the main, theoretical contribu-
tion which LSA tries to make. It looks at test construction and scaling
only as a special case of the larger problem of latent classifications
derived from manifest qualitative data. In the last analysis it aims at
attacking the broad issue of the relations between concept formation
and empirical research in the behavioral sciences.

REFERENCES

1. Aitken, A. C. Determinants and matrices. New York: Interscience,
1954. P. 9.

2. Allport, G. W. What is a trait of personality. /. abnorm. soc. PsychoL,
1931, 25, 368-372.

3. Anderson, T. W., Jr. An estimation of parameters in latent structure
analysis. Psychometrika, 1954, 19, 1-10.

4. Barton, A. H. The concept of property-space in social research. In
P. F. Lazarsfeld & M. Rosenberg (Eds.), The language of social research.
Glencoe, 111.: Free Press, 1955. Pp. 40ff.

5. Carnap, R. Testability and meaning. Phil Scl, 1936, 3, 419-471;
1937, 4, 1-40. Reprinted (abridged) in H. Feigl & M. Brodbeck (Eds.),
Readings in the philosophy of science. New York: Appleton-Century-Crofts,
1953. Pp. 47ff.

6. Dewey, J. Human nature and conduct. New York: Modern Library,
1930. P. 49.

7. Feller, W. An introduction to probability theory and its applications.
New York: Wiley, 1950. Pp. 214ff.

8. Gilliam, Sylvia. The latent structure of a T hurst one-C have scale. Un-
published Master's essay, Columbia Univer., 1948.

9. Green, B. F. Latent structure analysis and its relation to factor anal-
ysis. /. Amer. statist. Ass.3 1952, 43, 71-76.

10. Hempel, C. G. Fundamentals of concept formation in empirical
science. Int. Encycl. unified ScL, 1952, 2, No. 7.

11. James, W. The meaning of truth. New York: Longmans, Green,
1914. P. 149.

12. Kaplan, A. Definition and specification of meaning. /. Phil.., 1946,
43 (11), 281-288.

13. Koch, S. Clark L. Hull. In Estes, W. K., et al., Modern learning
theory. New York: Appleton-Gentury-Crofts, 1954. Pp. 67ff.

14. Lazarsfeld, P. F. (Ed.) Mathematical thinking in the social sci-
ences. Glencoe, 111.: Free Press, 1954. P. 377.

Latent Structure Analysis 543

15. Lazarsfeid, P. F., Berelson, B., & Gaudet, H. The peoples choice,
(2d ed.) New York: Columbia Univer. Press, 1948. Pp. 45-49.

16. Lazarsfeid. P. F. Interpretation of statistical relations as a research
operation. In P. F. Lazarsfeld & M. Rosenberg (Eds.), The language of
social research. Glencoe, 111.: Free Press, 1955. Pp. 115fL

17. Lord, F. The relation of test score to the trait underlying the test.
Educ. psychoL Measmt, 1953, 13, 517-549.

18. McClelland, D. C. Personality. New York: Sloane, 1951. P. 202.

19. McHugh, R. B. Efficient estimation and local identification in
latent class analysis. Psychometrika, 1956, 21, 331-347.

20. McRae, D., Jr. An exponential model for assessing four-fold tables.
Sociometry, 1956, 19, 2, 25.

21. Nagel, E. Principles of the theory of probability. Int. EncycL uni-
fied Sci., 1939, 1, No. 6, p. 23.

22. Rand Monograph No. 1.

23. Rand Monograph No. 2.

24. Rossi, P. H. The application of latent structure analysis to the
empirical study of social stratification. Unpublished doctoral dissertation,
Columbia Univer., 1951.

25. Somers, R. H. The latent content model. Unpublished Master's
essay, Columbia Univer.., 1956.

26. Stouffer, S. A., et al. Measurement and prediction. Princeton, N.J.:
Princeton Univer. Press, 1950. Chap. 1.

27. Thurstone, L. L., & Chave, E. J. The measurement of attitude.
Chicago: Univer. of Chicago Press, 1929.

28. Tolman, E. C. A psychological model. In T. Parsons & E. Shils
(Eds.), Toward a general theory of action. Cambridge, Mass.: Harvard
Univer. Press, 1952.

29. Tolman, E. C. The intervening variable. In M. Marx (Ed.), Psy-
chological theory. New York: Macmillan, 1951. Pp. 94-95.

30. Warner, W. L., et al. Social class in America. Chicago: Science Re-
search Associates, 1949.

31. Wiggins, L. M. Latent process analysis. Unpublished doctoral dis-
sertation, Columbia Univer., 1955.

WORK-EMOTIONALITY THEORY
OF THE GROUP AS ORGANISM

HERBERT A. THELEN

University of Chicago

Introduction 545

Background Factors {1} 545

Research strategy 545

Current approaches 547

Common elements 549

Research on work-emotionality 555

Orienting Attitudes {1} 557

Prediction 557

Level of analysis 560

Models 562

Comprehensiveness of empirical reference 564

Degree and mode of quantitative and mensurational specificity . . . 565

Formal organization of the system 568

Structure of the System as Thus Far Developed {2} 569

The nature of the groups studied 569

The development of postulates to guide research 571

Propositions about the nature of phenomena being studied 572

Propositions about human behavior in general 572

Propositions about the "group" 574

The research tasks 576

The sequential method for analysis of group operation 577

The measurement of valence and individual-group relationships . . 581

The identification of functional subgroups 583

Demonstration of relations to "productivity" 584

From private to public domains: "Blind analysis" and theory construc-
tion 585

The systematic independent, intervening, and dependent variables . . 587

Mode of definition of representative variables of each category . „ . 592

Major interrelations among constructs 594

Initial Evidential Grounds for Assumptions of System {3} 595

Construction of Function Forms {4} 598

544

Work-emotionality Theory of the Group as Organism 545

Mensurational and Quantificational Procedures {5}- 600

Formal Organization of the System {6} 601

Scope or Range of Application of the System {?> 602

History of System to Date In Mediating Research {8} 604

Evidence for the System {9} 607

Extensibility of Methods and Concepts, Programmaticiry, and Strategy for De-
velopment of the System {10-12} 607

References 608

INTRODUCTION

The basic concepts of the theory were suggested and explored by
W. R. Bion [3-9], then of Tavistock Institute, in a series of seven articles
called Experiences in Groups. Bion's concepts gave the research discussed
here its starting point.1 During 1951-1955, various technical reports
have presented our results to the Group Psychology Branch of the Office
of Naval Research. The major corpus of method and findings is available
in two monographs: one deals with methods [54], the other is concerned
with theory and findings [44]. The work was also presented in 1952 at
an American Psychological Association symposium [55], and it is dis-
cussed in The State of the Social Sciences [52].

BACKGROUND FACTORS

Research strategy. W. R. Bion [3] developed the concepts of work
and emotionality. Our work attempts to take the next steps of opera-
tional definition, refinement, and prediction.

The development of a body of principles requires the contributions
of men of different temperaments who can contribute what is needed

1 At nearly all stages the work has been discussed and thought about by the
research team. Nevertheless, major responsibilities have been accepted by indi-
viduals and subgroups. The team as a whole is mainly responsible for the sequen-
tial analysis method, although previous explorations had been conducted by deHaan
[13]. Ben-Zeev [2] alone developed the system of unitization, however. The develop-
ment and validation of the sentence-completion test was primarily the respon-
sibility of Stock, I. Gradolph, and P. Gradolph [54]; the first explorations, in 1949
were by Rosenthal and Soskin. The method for identifying functional subgroups
was primarily the responsibility of Stock and Hill [20, 54], with the help of
Stephenson [42] who largely developed the Q-sort method. The various studies of
productivity were conducted by different people: comparison of flight and work-
pairing groups, by I. Gradolph and P. Gradolph [44]; problem solving of 50 groups
was studied by Glide well [18, 44]; comparison of the two training groups was by a
team headed by I. Gradolph [44]; and the study of "trainability" was by Mathis
[33, 44]. Several further studies of individual-group relations are in progress by
Liebermann [31]. These and other subsidiary investigations, with full references
to prior studies, will be found in the monographs [44, 54].

546 HERBERT A. THELEN

at each step. There is probably a range of involvement within which
each of us works best. We need to reach out to the man with intuition to
help us conceptualize the content of the experience we wish to study,
and to the formalist or mathematician to see how to systematize and
give form and elegance to our body of intermediate principles.

The sequential, cooperative development may be illustrated by three
doctoral studies from the Human Dynamics Laboratory. The first [56]
set out to try to hypothesize basic categories of teacher behavior in the
classroom. The first step was to observe a couple of hundred hours of
classroom interaction. This was recorded by sound and by time-lapse
photography. These two records together enabled the observer, Withall,
to obtain almost perfect recall of his experience. Hour after hour he
spent studying the records. For each comment of the teacher, he asked
himself: "What, in my opinion, is the teacher communicating to the
class?" This is a subjective judgment. "If my judgment is correct," the
observer inquired then, "What sort of behavior should the students now
engage in?" This called for prediction of a general class of behaviors,
and this prediction had to be made from previously learned principles.
Thus if the teacher is judged to be making a punitive remark, then the
students should in general act like students who felt they were being
punished.

On the basis of this kind of substantially inductive thinking, Withall
arrived at 24 categories. These were then discussed by the research
seminar, and through more precise definition and theoretical examina-
tion, reduced to 6 basic categories so defined as to refer to distinguishably
different psychological motivations of the teacher. (We might have
trained ourselves to use all 24 categories, make many observations, and
treat the data with factor analysis; but it seemed more economical to
treat the matter conceptually — especially since there was already a body
of theory to go on.) Withall tested the applicability of his categories in
a couple of simple experiments — enough to persuade us that these cate-
gories were significant in the sense of being independent and interpret-
able.

Flanders [15] took the second step. He set up a simple laboratory
experiment, replicated seven times, in which he used the Withall cate-
gories as independent variables and then predicted consequences in terms
of a wide range of dependent variables. In other words, he trained
teachers to produce two quite different "styles" of teaching, as judged by
the Withall categories. Then he had them teach in an experimental situ-
ation and measured the consequences of the differences of the two styles
with respect to a variety of physiological, recall, achievement, and af-
fective variables. The fact that differences were found in the predicted
directions supported the notion that the independent variables actually

Work-emotionality Theory of the Group as Organism 54?

did index the psychological mechanisms used for prediction and, further,
that these mechanisms were properly conceptualized.

The third step was one of further demonstration that in a typical
everyday situation the same dynamics would occur. Perkins [36] selected
six classes of adults (all the preceding work had been with children)
and assigned teachers to them. Three of the teachers had been observed
to have one clear-cut style as measured by WithalTs categories; the other
three had styles similar among themselves but very different from that of
the first three. The dependent variables were confined to categories of
verbal response from members of the class. The experimenter collected
data from 10 meetings of each of the 6 groups, ending with some 15,000
responses. These wrere rated in 30 carefully defined categories, and
predictions were made as to the differences to be found in frequencies of
each category as a result of the two different styles. Twenty-one of the
predictions were borne out.

Thus we may represent scientific development in the study of such
problems as, in general, beginning with inductive-Intuitive hypotheses
obtained from firsthand experience with the phenomena. Then, in a re-
stricted situation, we have intensive study of a wide range of conse-
quences. Finally the hypotheses are demonstrated, using a narrower
range of variables but in a much more extensive situation, to make sense
in terms of "real" groups doing their regular work.

All that we require to minimize disputes over methodology is (a) a
large over-all view of the strategy of scientific development, (6) ac-
curate representation of where our particular work fits into this larger
strategy, (c) constant effort to reach out to make connections between
our work and that of others. The values animating the study reported on
in these volumes provide a further illustration of such a point of view.

Current approaches. Researchers operating with differing various
approaches to the problem will be found at some 15 or 20 "centers" for
group study. Each of these centers has its own goals, traditions, cohesive-
ness, standards, degree of individual freedom, differentiated roles, etc.
Although their communication varies both in volume and quality, by
and large, most researchers would probably agree that the work of the
various centers is complementary and overlapping.

Studies of groups have been made on many levels of complexity,
comprehensiveness, sophistication, and penetration, for different re-
searchers develop characteristic methodological and conceptual ap-
proaches [see 11 and 19 for attempts to sample major recent work in the
field]. To give some idea of the more prominent among these, I shall
follow the formulation of Cartwright and Zander [11].

Basic dimensions. Cattell and his associates use factor analysis to
determine the "major dimensions of groups." The factors are developed

548 HERBERT A. THELEN

from three kinds of variables: (a) measurable characteristic of individ-
uals, (b) structural characteristics of the group: the system of positional
interrelations shown in rank orders along various status continua, (c)
syntality variables, representing the performance of the group as a whole.
The dimensions represent, for the three kinds of variables, stable pattern-
ings in which, for different groups in different situations, the same vari-
ables enter into the same dimensions, but with different loadings.

Interaction. Bales, Homans, Chappie, White, and Arensberg are
considered as having a common interest in the way the group develops
and changes as the result of interactions among members and between
the group and its environment. These two kinds of interactions occur
in the "internal" system and the "external" system respectively. Bales's
theory relates patternings of 12 behavior categories to the processes of
developing group structure. Homans shows how the complex of activities,
interactions, and sentiments develops the internal and external systems
in their relationships to each other and to environmental factors.

Organizational leadership. Stogdill, Shartle, and Hemphill consider
the organizational aspects of group life: the functions and responsibilities
of individuals with reference to achievement of group goals. There are
both formal and informal networks of relationships among these factors.
The formal network defines expectations of role performance. The in-
formal network defines actual role performance. Leadership influences
both networks and finds problems in the discrepancies between them.
Leadership is understood in operational terms as an aspect of work per-
formance, work methods, and working relationships.

Psychoanalytic. Scheidlinger, Slavson, Redl, Bion, and Ezriel are
seen working with Freud's notion that "group cohesiveness arises through
common identifications of the members with one another." The meaning
of a particular behavior to the actor has to be understood both as fitting
into his genetically developed mode of adjustment and as involving him
in present reactions to the external world. "Personality" is the habitual
mode of synthesizing into a pattern of adjustment the aims of drives,
conscience, and physical and social reality (environment). Personality
develops through social interaction, especially in the family, and there
are discernible phases in its development. Opportunity for needed social
interaction is found in groups, and these form through common identifi-
cations, such as with the leader. Redl has identified 10 types of "central"
people with whom, under various conditions, members of the group may
identify and thus maintain the group. According to Bion's notion of
"valency," at different times unconscious subgroupings form through
"combination" in support of a particular emotionalized mode of group
operation. The group ethos is the organizing principle through which
individual strivings are coordinated in common effort. Role differentia-

Work-emotionality Theory of the Group as Organism 549

tions and social structure are produced in accordance with the capacities
of the individuals for interaction in the group.

Sociometric. Moreno, Jennings, Barker, Criswell, and others postulate
that the "social space" within which an individual lives is delimited by
his range of interaction with others and that this space is structured by
his feelings of attraction or repulsion for others. Such feeling bonds are
the bases of groups. Groups may be formed spontaneously from free
choices of others (psychogroup) or they may be formed by social de-
mands that require people to work together (soclogroup). The work-
ing out of interpersonal needs, as reflected in the choice pattern, is a
major aspect of group process, and the choice pattern at any time re-
veals significant cleavages, subgroups, and group structure in general.

Force field. Kurt Lewin [28], often considered the founder of "group
dynamics," saw that behavior arises out of the "life space" of the in-
dividual (cf. "social space," above). The life space contains percep-
tions of behavioral alternatives, such as different activities in which the
individual might engage. Some alternatives are definite and clear
(structured), others vague and unstructured. The alternatives have
different degrees of attractiveness and repulsion (valence) depending
upon their usefulness in meeting current needs. In addition, there may
be permeable or impermeable "barriers" to "locomotion" into the chosen
activity region, and these barriers also have a negative or positive valence.
Thus the individual, represented as a point within his life space, is sub-
ject to a variety of forces which tend to influence him in a variety of
directions. Applied to groups, these concepts lead to the notion that the
distribution of leadership depends upon the degree of overlap or com-
munality of the individual life spaces; cooperation and competition are
viewed [e.g., by Deutch, 14] as conditions under which the efforts of an
individual to locomote into a chosen activity region either facilitate or
hinder similar efforts by other individuals. The primary data for studies
within this frame of reference have been perceptions by members of
themselves, each other, their group, and its activities.

Common elements. One can sense many similarities in the assump-
tions which seem to underlie the various approaches. Although no single
rigorous integrated system can yet be set up, we can see dimly what sort
of propositions will probably be required. System building is a series of
mental operations, and we shall attempt to show how the various ele-
ments of the system emerge from successive and different operations.
While doing this, we shall also attempt to present in relatively summary
form — and with considerable diffidence — a set of theoretical elements
which may represent a composite view of the various approaches.

All living organisms engage in activity; and all living organisms have
organization in the sense of coordinated functioning of the different

550 HERBERT A. THELEN

parts. When we watch an organism, be it an insect or group, we can
collect three kinds of information: (a] we can describe the obviously
different physical parts, such as legs or members; (fe) we can describe
what the parts do — their overt behaviors; and (c) we can describe to
some extent how the whole organism moves : the insect toward a puddle
of water, or the group through an agenda.

On more careful examination, we may note that there are three
kinds of overt behaviors: (a) individual, as when a leg twitches or a
member fidgits; (6) interpersonal, as when two legs tangle or two mem-
bers argue; and (c) member-environment, as when the leg pushes the
ground or a member proposes a solution to a problem.

Over time, we become aware of regularities or consistencies between
behavior of each part and particular states of the organism and of the
environment. We draw inferences from comparison of the behavior of
the leg or member at different times and in different situations; and we
speculate about "what" might account for these inferred relationships.

At the heart of all our efforts to "explain" is the concept of purpose.
Other terms having the same kind of usefulness are need, drive, motiva-
tion, and tension (to be reduced). We "explain" by saying that the in-
sect or group or leg or member acts "as if it were trying to go some-
where, accomplish something, or deal with some condition external to
itself. We then infer that different parts (legs and mouth, Joe and Mary)
participate differently in accomplishing a particular purpose (satisfy a
need, reduce a drive or tension). We say that they have different func-
tions or take different roles and that the functions and roles are coordi-
nated in response to the over-all purpose of the organism.

At this point, however, our insect and group part company. There
are, after all, some differences between physiology and psychology. For
the question arises: why do the parts act together but differentially to
accomplish a purpose of the organism? And, more generally, how 'does
it maintain itself as an organism? This is the thorny problem of part-
whole relationships.

The differences between insect and group lie in the distribution of
sinews and nervous system. In the insect, the parts are connected physi-
cally: one nervous system coordinates movements; one brain directs the
whole enterprise. In the group, however, the parts are not connected
physically: the musculature is distributed among the members; instead
of one central directing agent there are as many agents as there are
members. Each member must then choose to belong to the group and to
participate under particular conditions and in particular ways. And this
process of choosing has both conscious and unconscious aspects.

At the conscious or nearly conscious level, we may take the utili-
tarian approach, saying that a member finds the group or activity at-

Work-emotionality Theory of the Group as Organism 551

tractive either because he expects to get some persona! reward from his
participation or because he wants to avoid some perceived possibility of
punishment. The reward might take the form of feeling better about
himself, some cherished person, or some valued idea; reward might be a
set of anticipations of new or additional opportunities, it might be what
he perceives to be the better opinion of others, etc. Similarly, punishment
might be perceived as coming from himself, others, or outsiders. In gen-
eral, we tend to relate reward and punishment in the group to relative
position : to be rewarded is to move into a more central position, to be
punished is to be pushed toward the outside. The concept of centrality
is related to esteem of the group, influence, personal adequacy in the
group situation, support of other people.

At the unconscious level, the key concept is the notion of identifica-
tion with others. This permits several possibilities. Everybody may feel
a tie to some one "central person," and through this communality, in-
dividuals identify with each other [37]. Everybody might have a com-
mon interest, in the sense of being attracted toward the same activity
(perceived in overlapping parts of each person's life space) [29]. There
may be no central object-tie at all but simply a web of interpersonal at-
traction, in which the persons mutually choosing each other form over-
lapping subgroups [23]. Again, the formative bond may represent a
tendency to act with others who are felt to desire the same emotionalized
mode of operation in the group [8]. But, whatever the tie that binds,
these various networks are perceived as belonging to the group as a
whole; the body of agreements, values, common perceptions, and com-
mon expectations developed through communication while working to-
gether constitute the culture of the group. This culture, including the
shared self-concept of the group as a unitary whole, becomes in itself an
object to promote intermember and member-group identification
(loyalty).

Included in the culture of the group is a set of goals with respect to
its environment. If we can say that the opportunity for individuals to
satisfy personality needs is the psychological raison d'etre for the group,
then we may also add that the common desire to change the environ-
ment (either social or physical) in some way is the social raison d'etre
for the group. Thus the group maintains itself for two quite different
kinds of reasons, and the behaviors in response to these two group needs
are not necessarily compatible. The psychological purposes are best
served by being a psyche-group, acting voluntarily on the basis of in-
ternal demands; the social purposes are best served by being a socio-
group, organizing to meet demands perceived as coming from outside
(from the nature of the problem, from the orders of a higher institutional
authority, etc.). This notion of the dual nature of the group is expressed

552 HERBERT A. THELEN

in various ways in different theories. We have mentioned the psyche-
socio-group distinction [24]. In addition, we can speak of the "internal"
system and the "external33 system [21], the "informal53 organization and
the "formal" organization [46], the aims of drives and the aims of reality
[41], the tendencies toward emotionality and the tendencies toward work
[5]. These dualities are not equivalent; i.e., they are not synonymous
words referring to the same thing. They are, however, different ways of
conceptualizing the dual nature of the group as being simultaneously
inner- and outer-directed.

We can now see that the fundamental problem of the group is to
satisfy simultaneously both sides of its nature, to meet individual needs
and to solve problems (change the environment). The group thus medi-
ates between the psychological, inner-directed needs of individuals and
the demands of the environment (as perceived by individuals), i.e.,
their social and environmental, outer-directed needs.

But when we say the group does something or the group has a need,
what do we mean? Have we now invented some superobject, the group,
and given it the biological properties of the insect? In operational terms,
this is in some ways the most difficult question of all. The problem boils
down to this: the member believes there is a group. He feels pressures,
expectations, and punishments, and he says they come from the "group."
He feels a need to "belong/5 not in the specific sense of having relation-
ship to particular individuals, but in a broader sense of feeling part of a
larger whole. He has a sense of place in this larger whole, and usually
he even can state his place in it. When talking to nonmembers, he re-
fers to the group as a unitary body, which has purposes, agenda, atti-
tudes, leadership, and so on; and he defends this body from attack. In
the mind of the member, then, there is no doubt that the group exists
and that it does have properties analogous to the biological. The "bio-
logical" group exists because the members think it does, because they
assume the others also think so, and because their behavior is different
as a result of these beliefs. And what makes a difference is "real."

The objective observer — if we may invent an idealized role that does
not exist — is baffled by the group. All he can put down on his record is
that Jim said something, then Jerry, and so on. To such an observer,
only overt behaviors exist; some one person is the actor and some other
person — usually the one who talks next — responds with further action.
And that is it.

The interpretative observer — and this is a role that does exist — goes
beyond the objective observer in his selection of "units33 to observe. He
still tallies individual behaviors, but he conceives of these in sequences
and periods of time. During one such period, the behaviors may indicate
to him that the participants are confused: they make suggestions but

Work-emotionality Theory of the Group as Organism 553

do not follow them up; they express a range of feelings, from depend-
ency to aggression. During the next period, however, the participants
seem to be listening to each other. They summarize their discussion;
they ask each other questions; they look comfortable; their feelings are
warm. It is as if the same thing had "gotten into" all of the participants
at the same time, and this is different from whatever was bothering them
earlier. This "thing" does not "cause35 all the participants to behave the
same way: there is here no Greek chorus chanting in unison; each person
is still recognizable as the same person in both periods; each person in
many ways is behaving differently from each other person. Yet there is
something to be accounted for that affects them all; and since all the
participants are affected by it, this thing must be communal.

What about the nonparticipants? Are they also affected by this
communal thing? Our cautious observer wrould have to admit (a] he
has no data on the nonparticipants except possibly for some nonverbal
gesturing which is difficult to interpret; (6) people cannot all talk at
once, and the selection of those who do talk and the inhibition of those
who do not may be opposite sides of this same communal "thing53; (c)
during the next period, these nonparticipants become the talkers, and
the best hunch is that they wrere getting ready for this during their pre-
ceding silence. In other words, not talking is not nonparticipation.

The problem now becomes: what is this communal thing? Two
points stand out: it is probably something that exists in some way in the
minds or nervous systems of the members; it develops and changes
through processes of interaction and communication among them. More-
over, since this common thing seems to affect ideas, emotions, actions,
and values, it is a total pattern which contains either these things itself
or the anlage or precursor of them.

All the varied theoretical approaches must deal with this problem in
some way. The cultural anthropologist, studying the group much as he
would study a South Sea tribe, may find the common thing in basic as-
sumptions or organizing principles in the culture, and these are the key
ideas through which he understands the "way of life." To a "structural"
sociologist, the common thing may be the "shared" expectations mem-
bers have for each other's behavior or roles. To the "functional" sociolo-
gist, this common thing may be shared interests and purposes. To the field
psychologist, this common thing may be an imbalance of "forces," with
strains and stresses to which all are responsive. To the psychoanalytically
oriented observer, this common thing may be an "internalization" of the
group as a common conscience, or as an extension of the individual's ego.
To some extent, most researchers find themselves consciously emphasiz-
ing one of these notions, but continually assuming additional elements
from other approaches as well.

554 HERBERT A. THELEN

Perhaps we can conclude only that a group is a number of people
who think they are a group and act like one. Nevertheless, a "group"
is a real thing, even though it is a theoretical construct rather than a
physical object. Hereafter, then, we shall use the word "group" to stand
for the common "thing" in any or all of the senses just indicated.

The dual nature of the group — what we called its inner- and outer-
directed systems — means the possibility of conflict, tension, ambivalence,
and ambiguity; and it also means such reactions as anxiety to these
states. Moreover, it means efforts to resolve, reconcile, or harmonize
these warring elements. The group needs to do this, not only because
these states are painful and punishing on the whole, but also because
people need the security that comes from a defined situation. In addition,
they need a tolerably orderly society so that they can employ their intelli-
gence to guide their behaviors in the light of predictable consequences.

But efforts to this end only spell out some further ramifications of
the problem of maintaining the group. The familiar devices are em-
ployed: there are shifts in central people, changes in group structure,
redefinitions of purposes, activation of different, hitherto merely po-
tential individual needs, shifts in the formal organization, new feelings of
intermember attraction and repulsion, and so on. Out of this come
changes or reorganization of the culture (usually lagging somewhat be-
hind) ; and the group moves into the next phase.

Some of these components of change are consciously guided; others
occur at "deeper" levels. In so far as man is master of his fate, it is
through the conscious use of intelligence; hence, he must either work
directly on the elements in consciousness or he must consciously set up
conditions in such a way that troublesome elements can emerge into con-
sciousness. This latter method centers around removing the obstacles,
such as personal threat and fear, that tend to keep disturbing elements
suppressed. Elements for the most part already in consciousness, and
which can be dealt with by means easily available, are the objective con-
ditions usually seen as lying "outside" the group. In other words, in-
telligence may be more easily directed to problem solving or if you like,
to achieving the group's publicly stated purposes or tasks. Moreover,
this task-activity, with its clear-cut requirements of information to be
secured, roles to be played, methods to be used, provides a set of "givens"
against which the group can diagnose difficulties and evaluate and rectify
its internal conditions. Thus work on tasks provides both "feedback" to
the group about its own adequacy as a group and a clear-cut target for
channeling expressions of personal feelings. The progress of problem
solving, together with the attendant need frustrations or satisfactions,
is something each person can judge independently and have opinions
about — and this is the grist for the problem-solving mill. Thus in the

Work-emotionality Theory of the Group as Organism 555

group's conscious operations, the outside demands for particular ways of
structuring the group tend to be perceived as fundamental; and the in-
ternal structure must accommodate to these.

At the unconscious level, however, the situation Is reversed: it Is the
process problems connected with the inner system that most affect moti-
vation, identifications, "place," and so on. The problem-solving opera-
tion merely provides the vehicle for these higher-priority "emotionalized"
human purposes [50]. In operations at the deeper level, there Is no
doubt that different individuals become "central33 from time to time;
Bion [8] has even described one situation in which an absent member
was central.

It Is at this point wre must leave the group. We have attempted to
show what the researchers are working with and to Indicate what seem
to us to be rather generally held or assumed notions about groups. Many
things have been omitted, e.g. overlapping memberships, genetic devel-
opment, group growth, and so on; but perhaps we have Indicated the
central problems most directly Involved in group operation as well as
some major concepts currently being applied to these problems.

Research on work-emotionality. We shall introduce the particular
research [44, 54] which Is the subject of this paper later, under Structure
of the System as Thus Far Developed. But it does seem appropriate at
this point to note briefly where it fits within the methodological and con-
ceptual framework which we have sketched so far.

The basic dramatic theme is that proposed by Bion: the group Is
seen as a miniature society which has conflicts within Itself over the basic
assumptions on which it operates. The two general types of assumptions
refer to different major purposes, to "work," in the sense of dealing with
reality factors diagnosed as creating problems to be solved, and to "emo-
tionality," which attempts to avoid certain reality factors but which at
the same time serves to help maintain the group.

The structure of the group as a series of status hierarchies is dealt
with only when the members are themselves actively concerned with
problems from this source. The structure of the group as emotional in-
terpersonal alignments comes into the picture as necessary to explain
participation or inhibition or interpersonal conflict. The structure of the
group as functional relationships conies in to the extent that the need for
definition or modification of such relationships influences group activity.

The basis of group formation and groupness In general is "combina-
tion55 [8] with others in the support of particular basic assumptions. Thus
the group is seen as a shifting network of actual and potential subgroups
which changes in response to the altered needs of the group.

Individual needs are understood as necessary to explain the ways in-
dividuals participate to influence the basic assumptions operating during

556 HERBERT A. THELEN

each phase of group life. The conditions within the group as a whole
activate each person's behavior. Although individual behavior is an ex-
pression of individual need, the conditions under which it arises and the
way in which it is expressed is symptomatic of group need. In this sense,
the individual is always to some extent speaking for the needs of the
group.

The group is seen as working simultaneously on task and process
problems and as providing the conditions for individuals to wrork on
unique personal problems. Task problems are equivalent to external
problems or problems in the external structure, but they come from needs
arising from within the group and are projected into the environment
(externalized) by action of the group itself. In other words, environ-
mental factors are brought into the group by each individual and they
eventuate in group tasks through processes of opinion exchange and
group decision. Process problems arise in the efforts of the group to or-
ganize itself to work on task problems and in its efforts to maintain
itself as a group. The task problems provide the necessary vehicle and
frame of reference for processes of group maintenance. By maintaining
the group we mean developing and stabilizing during each phase a
situation of dynamic interplay such that the most imperative needs of
individuals can be met.

Methodologically, these notions lead to observation of group process
as the source of the most fundamental data. Behavior is seen as resulting
from the momentary mixture of tendencies toward work and toward
emotionality. These are recorded in categories developed for the purpose.
The observer is emotionally involved, and identifies himself with the
group as a whole, its sense of conflict, tension, and so on. In effect, he
records the contribution of members to the prevailing emotion-work
"climate." Emotion is directly felt and recorded as it is expressed; it is
not introduced later as a theory to account for "objective35 behavior.
The observed categories of each member's contribution are plotted as a
sequence in time, and this sequence is divided into "natural units" or
phases.

During each phase, the task and process problems are identified.
The task problem is usually explicit; the implicit process problem is diag-
nosed through interpretation. Both formal (task) and hidden (process)
agendas of the group during a phase, a meeting, or during its whole
life are thus revealed.

The explanation of these group-level phenomena is attempted
through use of a second kind of information, namely, the tendencies of
individuals (in group situations) to support or oppose each of the pos-
sible basic assumptions and also the tendencies of individuals to react
in defined alternative ways to such assumptions when they exist in the

Work-emotionality Theory of the Group as Organism 557

group. Tests have been devised for obtaining this information. We have
also attempted to explain the group-level phenomena as the interplay
between subgroups representing "combinations35 of individuals for the
support or inhibition of particular basic assumptions. These subgroups
are identified through factor analysis of self-perceptual items arranged
in a Q sort. Some of the subgroups tend to correspond to sociometric
subgroups, whereas others do not.

Members' own reports of their reactions to events in group life (e.g.,
postmeeting questionnaires) are used at points where we are interested
in their own theories about what is going on. Such reports are never
taken at "face33 value; they must be interpreted in light of other data
about the needs of individuals in the situations on which they are re-
porting reactions.

In general, there is a close parallel between our methods of research
and those of a somewhat psychiatricafly oriented group-centered leader
or member. The research group has found these ideas helpful in the
practical business of leading groups (especially those for training in
human relations ), and the hunches of group leaders have been found
to be rather readily translatable into hypotheses for study within this
framework of concepts and methods.

ORIENTING ATTITUDES

Prediction. Research involves theorizing and reality testing. By
theorizing, I mean formulating and systematizing ideas which "explain55
our experiences. By reality testing, I mean demonstrating that concep-
tual relationships correspond to behavior in the "real" (i.e. nonsubjec-
tive) world. It is important to accept the fact that theorizing is a way of
meeting the experimenter's need for "closure," for "wrapping up35 parts
of his own experience.

Reality testing, on the other hand, is the effort to externalize ideas
hitherto in the researcher's subjective world by showing that they "fit"
the experiences of other people as well. The pursuit of ideas, necessarily
individual and self-centered, becomes socialized through the socially
prescribed methods of testing hypotheses.

Thus the scientific enterprise represents a movement from subjectivity
to objectivity. If an idea survives after being squeezed through the sci-
entific wringer, then it is entitled to be accepted as a present approxi-
mation of the "truth" within a carefully described field of designated
phenomena. Such an idea will have currency until another, more fash-
ionable type of wringer that would tear the idea to shreds is perfected,
or until another idea, no more "true" but more convenient or useful,
comes along, or until the idea is incorporated in larger systems of ideas.

558 HERBERT A. THELEN

To carry through the two basic parts of research requires two quite
different roles: the researcher acting as a human being to get closure
from his own experience; and the researcher operating as a member of
a group of scientists in whose behalf he must submit his brain child to
cold test.

The crucial test is prediction. The classical question runs: given situ-
ation at time a, what will it be like at future time b? The operations re-
quired to answer this question are somewhat as follows: study situation
X, name its "parts" and describe the part-to-part and part-to-whole rela-
tionships within the situation : this is called determining the structure of
situation X. Next, study this structure, and ascertain that it contains (as
all human systems must) certain suspicious-looking stresses and strains:
this is diagnosis or analysis of growth tendencies as revealed sympto-
matically in present dislocations. Consider what sorts of rearrangements
of parts or changes within parts would reduce the stresses and strains:
this is the making of models which portray the system in more "stable
equilibrium." Then estimate the probability that each of these rearrange-
ments under the given conditions might occur. (This is equivalent to
describing the potentials for flow of "energy" in different directions
within the system, but the probability-states concept seems less likely to
embroil us in bad analogies to physical science.) Having selected the
most probable state, the next question is: how far will it have been
realized by time b? The answer to this question, preferably conveyed in
the form of a picture of the structure at time fe, is our prediction, and all
that remains is to wait until time b and then look for evidence that the
predicted and observed structures are alike.

Unfortunately, even successful carrying through of these steps is not
enough. The researcher must also be able to make explicit all the con-
ceptual relationships he uses and show us the relationships between these
and his data.

A comforting human fact is at the same time awkward for science:
the wisdom amassed from experience is greater than the contents of ex-
plicit knowledge. This makes it possible to short-circuit one or more of
the steps required by our idealized model. The successful researcher on
groups usually arcs across these steps In a flash of insight. This may make
him a bad scientist but an excellent companion for a seeker-after-hy-
potheses who knows a good thing when he sees it. Denied the easy and
"human" way out, the scientist may delimit the experimental situation
to a few measurable aspects whose relationships are clear-cut (as in
questionnaire studies of perceptions reported by group members). He
may limit his own aspiration to doing a thorough job with one step,
leaving the rest to others (as in "perfecting" a measuring instrument).
He may take on the whole job, repeating it over and over and gradually

Work-emotionality Theory of the Group as Organism 559

making the concepts more and more explicit as the whole framework
becomes clearer (as in developing a psychoanalytic theory of leadership" .
Or, finally, he may reconsider the model for prediction and reformulate
it to fit each situation.

To the extent that research is learning, the first three approaches
overlap. The perceptionist's experiments do not seem empty to him be-
cause he has an implicit frame of reference that enables him to con-
sider perceptual data as indexing more important underlying dynamics.
The test constructor likewise has some broader set of principles which
enable him to try to validate his instrument. And the developmentalist
must pay his respects to the others along the way. These conditions stem
from the facts of learning: that behaxior arises out of a broader sub-
jective world and that adaptation requires conscious contact with reality.
Nevertheless advance would come faster if we could free ourselves to
encourage the implicit to become explicit and if advocates of the various
approaches could be a little more confident of their need for coopera-
tion.

Probably, use of the final approach suggested will make the researcher
as learner most effective. Let us reconnoiter the possibilities of making
our experimental designs more suitable to the degree of fragility or case-
hardening of the researcher's concepts and intentions.

The possibilities may be set up logically through a simple analysis
of the number of kinds of relationships that can be developed with refer-
ence to a situation. We offer, at a rather high level of abstraction, one
such analysis to show what we mean.

1. Prediction may be made to something in the past, present, or
future, depending on whether one is concerned with "causes," correlates,
or "effects."

2. The "something" predicted from and/or to has the character of
either structure or process.

3. Structural variables may represent either wholes or parts of
structures.

4. Process variables may represent either events (homogeneous units
of interaction over time) or specific behaviors (seen as parts of events).

5. On this basis, there are 34 types of predictive statements. The
breakdown is

a. From a structure to one or more parts, an event, or one or more
specific behaviors.

6. From one or more parts to a structure, an event, or one or more
specific behaviors.

c. From an event to one or more behaviors, a structure, or one or
more parts.

560 HERBERT A. THELEN

d. From one or more specific behaviors to an event, a structure, or
one or more parts.

<?. This makes 12 possibilities.

/. Each possibility may refer to past, present, or future, with the ex-
ception that event-to-event and structure-to-structure predictions
(whole-to-the-same- whole) are identities rather than predictions.
Thus we have 34 types of prediction.

6. The purpose of prediction is to test instances of application of
general hypotheses. Hypotheses are statements of relationships between
parts and whole, whole and whole, and/or parts and parts. The rela-
tionship may indicate antecedence, simultaneity, or future consequence.
Thus there are as many forms of hypotheses as there are types of predic-
tive statements.

This perhaps suggests my attitude toward prediction: experimental-
ists should analyze concepts dimensionally to see what formal relation-
ships are involved in our hypotheses. Secondly, we should validate our
hypotheses by creating prediction situations (experimental designs) of
the appropriate types. Systematic development of our own metatheories
along these lines would enable us to convert the scientific wringer from
a fashionable gadget to a feedback device for improving and developing
ideas, regardless of their stage of maturity, fragility, or harmony with
current fashion.

These are the major types of predictive statements made by our own
research:

1. Prediction of present structure of personality as a whole from a
limited number of selected and quantified aspects (parts) .

2. Prediction of future functional role (participation pattern over a
designated period of time) from personality structure.

3. Prediction of present events (underlying group problems) from
observed specific behaviors.

4. Prediction of specific behaviors within future classes of events
(modes of group operation) from present personality structure.

5. Prediction of group structure (future whole) from present per-
sonality structures (parts).

6. Prediction of future events in group life (whole, modes of opera-
tion) from present group structure (whole) .

Level of analysis. Whether as structures or events, wholes are not
directly measurable. Wholes are theoretical constructs. By using them,
one can organize a large mass of specific information and give it co-
herence. If their internal organizing principle can be made explicit,
these constructs also can be used as bases of prediction. Organizing
principles within structures are such notions as the existence of differ-

Work-emotionality Theory of the Group as Organism 561

entiation of members along Interrelated dimensions of prestige, influence,
resistance, etc. More abstractly, the structure may be given coherence
through some concept of stability, scope, or extent '''e.g., life space, social
space). Organizing principles within events tend to portray bask tend-
encies to move In some direction; thus these principles are projected, "as-
if 3 purposes or dynamic themes as discussed earlier.

The concept of the whole must represent a much lower degree of
specificity than the data from the parts. Presumably the course of In-
vestigation Is designed to help us oscillate between data and constructs.
In this fashion, we move up and down the ladder of abstraction from
the observer's sensory responses to organizing Ideas which pattern these
responses, to specific responses in defined (controlled) situations, to
further patterns, and so on. If one looks at the design In terms of the
phenomena to which the researcher's behavior refers and presumably
corresponds, then the same movement is seen moving from parts to
whole to parts to whole.

This kind of movement is the essence of finding "meaning" because
these levels have different properties. The specific level Is descriptive;
the construct level is explanatory. And both description and explanation
are required for "meaning,55 i.e., for the kind of Internalized insight that
we call comprehension. Laws of human behavior can be found only
at the construct level; e.g., frustration leads to aggression. These are
"whole55 emotionalized states rather than specific behaviors. If you
ascertain that a man Is frustrated, you can predict not what he will do
specifically, but what sort of thing he will do. If you add further in-
formation about the sort of man this particular individual is — informa-
tion about his typical ways of dealing with stress situations — then you
can narrow the prediction to a few kinds of response tendencies typical
of this person. If to this you add information about specific aspects of
the situation, and if you have a theory about how the person selects
which of the possible kinds of response he will produce, then you can
predict his behavior more specifically. For completely specific prediction,
you would have to know all the behaviors in a sequence leading toward
the precise moment you are trying to predict to; and at this point pre-
diction and observation would be practically synonymous. In general,
the more particular the prediction one wishes to make, the more prin-
ciples he must know, the more specific the data he must have, and the
shorter the time over which he must attempt prediction.

In our own work, the basic categories are of behavioral tendencies
whether they be seen in personality (intrain dividual) or in observed
behaviors (interpersonal). We use eight fundamental categories; four
refer to "emotionality55 and four to C£work.55 Thus our conceptualizing
is at a rather high level of abstraction, with rather coarse instruments

562 HERBERT A. THELEN

as compared, say, to the use of up to 50 "traits" or 12 "dimensions."
Our description is at a very low level of abstraction: we record every
word spoken, and we rate ever)7 contribution of an individual.

Models. Behavior is purposive. The purposes may be conscious, as
in problem solving, or implied, as in the consistent effort of a person
to "dominate." Behavior is also part of a stream of experience and par-
ticular behavior is an emergent event within a larger personal-social-
physical system. Thus we speak of a particular behavior as "sympto-
matic" of underlying conditions or states. When we respond to a be-
havior of someone else, we are responding also to whatever the behavior
signifies to us.

Hence, any category for classifying behavior — particularly at a
relatively high level of abstraction — is likely to imply more than it
specifies. It is also likely to imply somewhat different things to different
users of the category. We acknowledge this when we work to avoid
using "color" words in our operational definitions, but what word does
not have at least a faint hue for someone? The fact that most descriptive
and most active words have color illustrates the difficulty of plucking an
overt act out of its context of relationships and sentiments.

Faced with this difficulty, the category maker has available two
courses of action. He may concentrate on physical description, which
he can do only by enumerating ah1 the possible instances of the category
and then using this as a check list. Secondly, he may describe the sort
of behavior he means, and then say what he means by it: under what
conditions it is likely to arise, how others may react to it, what may
be the apparent intention behind it, and so on. The first procedure
amounts to describing the physical and audible aspects of behavior; the
second amounts to describing action in a situation. Each has its own
drawbacks.

Thus, when we try to categorize behavior through its "overt"
aspects, we find that it is impossible to list all the instances of the
category. Many behaviors are not pure instances, although they have
some elements of the aspects we have listed. To list many instances
within a general category does require some high-level concepts. Fourthly,
if the observer has only physical aspects to guide him, then the con-
text may be too small for objectivity. He tends to fill out some of the
missing dimensions and then make the judgment fit; for if the cues one
is looking for are too minimal and unmeaningful, it is easy to overlook
or misperceive them. For example, we found that our observers agreed
more completely on the judgment of the amount of work per con-
tribution during a meeting than on the number of times each person
spoke. Yet the latter item is perfectly "objective" and the former re-
quires a relatively complex judgment.

Work-emotionality Theory of the Group as Organism 563

The difficulties of the second type of category, in which both the
behavior and Its context are spelled out, are first, that one cannot be
sure just which of the criteria were used to classify any particular be-
havior. The theory most be fairly fully developed before such "full"
categories can be provided, and, in that case, the category is not really
just a category of overt behavior: it is actually a hypothesized dynamism.
Finally, since the category does not denote particular beha\iors specifi-
cally as belonging or not belonging to It, the observer must do more than
observe and check : he must interpret as well.

When one tries to use categories of the first sort, they get perverted
into the second sort. That is, observers elaborate the simple description
with their own private meanings. They cannot succeed in attempts not to
respond to affect and intention. Endeavoring to cut off their involvement
in the phenomena leads either to psychological withdrawal of the ob-
server, which tends to produce carelessness, or to efforts to suppress and
deny involvement — which tends to distort perceptions. Hence, we have
preferred to fill out the context as part of the definition of the category,
and simply to accept the fact that we are not checking a specific be-
havior at all; what we are checking is a symptom of a general mode
of response.

This means that our categories for behavior are actually models of
different modes of response or action. They denote not only a class of
specific actions in a generally defined situation; they also connote "as-if"
purposes and expected consequences. Thus, building categories of "be-
havior" is actually a process of developing subtheories about dynamic
processes which are — and which must always be — the actual objects of
any investigation of human behavior.

We conclude, then, that observation is an activity which involves
thinking along with the group. The categories of observation, however
defined, become distinguishable "dynamics" or modes of adjustment or
accommodation. If we attempt to make the definition of the categories
correspond to the definitions we actually use, then our categories become
alternative models.

As we move to higher interpretative levels, we find increasing recog-
nition of the use of models. For the higher the degree of abstraction, the
more wholisticaUy we understand what is going on, especially in regard
to the social directions (as apart from the psychological dynamism) to-
ward which we think the system is moving.

As social creatures, at this point (if not earlier in some "moralistic"
sense), our evaluative reactions come into play. A scientist may claim
that he has no evaluative reactions to such things. But probably it is
more useful to recognize evaluative reactions and analyze them, for
the fact that we feel as we do provides another kind of datum for

564 HERBERT A. THELEN

analysis and another route to comprehension of the situation. Just as
we cannot avoid using models for activity, so we cannot avoid using
models for structures and systems of part- whole relationships. Words like
cooperation, competition, social order, freedom are not simple concepts;
they are foci for the association and organization of a great many
relationships, models toward or away from which we perceive the system
to be moving. It is the existence of these models, and the values we
impute to them, that causes us to have evaluative reactions in the first
place.

We regard our basic categories of emotionality and work as alternative
models for modes of adjustment and for states of group culture. We look
at the behavior, but we see it in terms of a situation and in terms of
probable drives and group needs.

Comprehensiveness of empirical reference. The study of a face-to-
face group is the study of interaction between personality' and culture.
The group is an intermediary body for bringing these two kinds of
phenomena into relationship. The group's mode of adjustment represents
a set of assumptions which govern the mediation process at any par-
ticular time. The study of groups is also the study of man and society;
events in group life illuminate the biological-psychological nature of man
and the sociological-anthropological values and organization of the
larger society. Hence, it is scarcely surprising that a goodly number of
the men investigating groups are also concerned with the possibilities of
moving toward a larger, all-embracing science of human behavior.

Our concepts have "surplus" meanings in terms of universals. The
concept of "dependency" as a mode of adjustment is comprehended in
a model which incorporates our ideas about the intentions and needs of a
dependent person along with ideas about the system of group control
and the group's goal directions ; these are related in the model to expecta-
tions about the immediate situation and to trends in the larger society.
No model is ever fully spelled out. Some of these ideas are explicitly
related, as in mathematical laws, but some stand at a verbal enumerative
generalization level, whereas others remain at various distances below
consciousness, waiting for the proper combination of cue and drive to
make them emerge into consciousness. The value of the model lies in the
very fact that it is full of surplus meanings; it implies potentially a
great deal more than it ever states. In my opinion, research is concerned
with discovering these surplus meanings, making them explicit, re-
organizing the model as needed to tighten its structure through penetra-
tion of its organizing principles, and, finally, using these principles for
prediction. The predictions that pay off at the 1 per cent level may
contribute most to the organized body of science (at least they are more
likely to be "accepted" ) ; but the 20 per cent level can be far more

Work-emotionality Theory of the Group as Organism 565

interesting. For It may show that we have correctly grasped the general
Idea but are overlooking factors which Invite Investigation although we
had not thought of them.

All of this means that the "comprehensiveness of empirical reference"
is potentially very broad. Thinking about groups embraces almost the
wrhole range of Ideas about Individuals and society. And every human
interaction — intrapersonal, Interpersonal, group, and community — pre-
sents the opportunity to draw on the complete range of Ideas. Thus
Romans [21] applies the same over-all model to a small face-to-face
group on the one hand and to a whole community on the other. After all,
science must assume that the whole world Is organized along the lines of
discoverable principles, and that each part, In some way or another, fits
In with these principles. Otherwise It would be Impossible to generalize
from the situation studied to other situations. In the early stages, when we
lack confidence in our general principles or, perhaps, In the language we
use to try to express these principles, it is reasonable to demand that
we restrict our generalizing to the cases we have actually studied. This
is the counsel of caution and the requirement of public demonstra-
tion, it is not the way we think nor does it express correctly the long-
range goals and the basic assumptions of science.

Degree and mode of quantitative and mensuratlonal specificity. The
problem of language is critical. We must communicate the central idea
of any concept by surrounding it with qualifying phrases, as ifs, ands, and
buts. Our grammar, with its simple subject-object relationships, and Its
adjective and adverbs to modify these relationships, seems better suited
to discuss objective phenomena than psychological events. We have
much more the feeling of "pulling phenomena apart," and of "reifying"
images like the image of a group when we operate in the human sphere.
Time relationships are still more difficult as are ideas of cause and effect.
Language may derive from the need to describe objects. It offers a
serviceable substitute for pointing, but in the realm of human behavior,
language may be less adequate. Thus Harry does not react to Joe. This
our language can handle easily. But it is apt to falter in dealing with
complexity. Instead of reacting to Joe, Harry reacts to Harry's feelings,
which are mobilized in a situation in which Joe is also present, as a
result of some behavior of Joe's which, in the context of the general
relationships of mutual expectation and attraction between Joe and
Harry, cues off the mobilization of feelings specific to the situation.

To meet problems of this sort, one may invent new words and try
to make their meanings stick. One seeks to develop a kind of shorthand
which, it is hoped, will be both more terse and more precise than the
phrases it renders obsolete. Or, one may use an entirely different kind
of language — the language of mathematics, Words are the language of

566 HERBERT A. THELEN

essences and entitles; mathematics is the language of relationships. In
the sense that we probably respond to behaviors and activities and then
try to define the entities that fit into these relationships, mathematics
seems more suitable. Thus, for example, we do not "know" Harry's per-
sonality. This is merely a theory to account for the fact that there are
certain consistencies in the way things happen when Harry is part of a
situation and that these consistencies of relationship between Harry and
other people, objects, values, etc., are located in Harry — or rather, in
Harry's personality. Harry thus is the locus of one end of an infinite
number of relationships; and it is the relationships we experience. Harry
is just a theorized entity — even if we can borrow money from him.

When we talk about relationships we are talking the language of
mathematics, and the mathematicians may have grounds for their con-
stant wonder why we do not go the whole way. But going the whole
way is not easy. It requires the development of a set of dimensions
which represent continua and which, taken together, enable us to build
even the most complicated variables. Thus in mechanics, the three
fundamental dimensions are distance, mass, and time. These are all
measurable along continua, and more complex variables such as inertia
and force are synthesized as unequivocal relationships among these
dimensions.

The effort to find univocal relationships implies a deterministic view
of behavior: if we could measure the magnitude of a number of chosen
variables, and if we knew the mathematical relationships of covariation
among them, then, given new values (as at another time) of any one of
them, the values of the others could be computed. And, what is more
important, they would fit the new facts in the situation. As scientists we
must believe in determinism; that is what science is about. But when we
deal with the human system there are such embarrassing factors as feed-
back, which modifies the system as a function of certain consequences of
its own operation; choice, which means that more than one alternative
route can be perceived; analysis and consciousness, which mean that
more alternatives can be felt and formulated, and so on. All these
processes must be reduced to lawful behavior and incorporated also into
the determinist's system. Finally there is that old friend of the statistician,
the "randomly distributed uncontrolled factor,35 which I tend to think of
as the ignorance factor, but which at least calls our attention to the point
of view that prediction is possible only within limits and that whatever
prediction is made should be accompanied with a probability tag.

The usefulness of mathematics as a language raises the funda-
mental question whether by adding more and more deterministic ele-
ments (including the determination of probabilities) one can ever ap-
proximate the facts of a human system. Certainly we can only expect

Work-emotionality Theory of the Group as Organism 567

the mathematicians to find more and more ways to remove from an}*
given event all the systematically varying elements, and to reduce to an
Increasingly manageable size the part of the event's determination which
cannot be put into formulas. But we have still to see how well Korzybski's
[25] notion that '"the structure of the universe and the structure of
mathematics Is the same" will stand up in human affairs.

We have said that we react to dynamisms which are constituted by
relationships among the behaviors involved; and that It is easier to sense
the relationships than to define the entities, e.g., Joe and Ham*, thus
related. It is easy to know that a group today seems to be behaving
very differently than It did yesterday but how do you measure a group
or group behavior? Group and group behavior are both constructs. This
means they must be synthesized from a host of smaller Interrelated
measurable elements. What then do we measure? One researcher meas-
ures perceptions of members, which he can then intercorrelate and factor-
analyze. Another counts the number of times he sees Instances of be-
havior of up to a dozen or more "types." A third researcher pays at-
tention to "group55 productivity., by which he means and measures such
things as the time needed to work a puzzle, or in Industry, the number
of relays assembled during successive specified periods. The "empirical"
approach thus bolls down to measuring everything you can and hoping
there will be relationships among these things; whereas the theoretical
approach involves formulating, almost by aesthetic criteria, the Idea of
different kinds of movements, trying to express them in words, and then
hoping that some way can be found to measure what one thinks he has
in mind. In our opinion, the best strategy involves both approaches:
constant effort to guide empirical trial and errors by ideas of sensed
relationships, and, when a measured relationship holds firm, effort to
see what it means within the larger theory.

In our own work, our measurements are frequencies of appearance
of different categories of behavior in test and group situations described
as well as we can describe them. Our measured quantities are only in-
dices for the most part — symptoms of a state of affairs which can be
comprehended only through theoretical reconstruction. The degree of
specificity depends upon the question we are investigating, which may
range from "under what conditions will Johnny participate in x type
of discussion" to "what changes occur in a training group over the course
of 15 meetings?" In both cases, the raw data are similar, but the col-
lation of the data, the use of sequential analysis and field graphs, and
the balance of qualitative to quantitative operations with the data — all
these depend upon the question.

In general, the specificity desired is obtained by taking into account
an appropriate number of indexed tendencies. Uniqueness represents, in

568 HERBERT A. THELEN

effect, the convergence of a large number of different tendencies at a
particular time and place.

Formal organization of the system. A researcher is a human being
engaged in inquiry. He must have a goal, a frame of reference, a means
of generating hypotheses from the frame of reference, and experimental
methods for testing his hypotheses. Finally, he must develop a body of
theory in which the terms are defined through their relationships to each
other, and for which the surviving hypotheses provide reality contact
adequate to support the whole system.

When this last or theory-building stage is accomplished, the re-
searcher has summarized his whole enterprise within a formally or-
ganized system. From such a system, he and others can "deduce"
many further hypotheses which further specify the system. If the con-
cepts in the theory were constructed from a limited number of opera-
tionally defined dimensions, then new concepts (and new theory) can
be added indefinitely through the systematic consideration of all the
possible ways of mathematically combining these dimensions — as in the
case of CGS system in physical science. At this point, the system might
be said to have attained relatively advanced "hypothetico-deductive"
status. (Whether this dimensional breakdown and build-up is actually
possible in social science, I leave in the realm of questioning. )

In viewing research as inquiry, we are implying that the researcher
guides himself consciously through the steps given earlier. Current fashion
may lump everything leading up to the statement of hypotheses as
"creating hypotheses" and regard this as the domain of private in-
spiration. Before we agree to this, however, we should study a bit more
fully the relationships between "frame of reference" and "hypotheses53 —
and also the relationships between physical and social science. It is pos-
sible that we have been too long and too much influenced by Kekule
who, so the story goes, became frustrated, withdrew to a beer garden,
drank too much, and, in a dream, saw six snakes grab each other's tails
in their mouths and thus form a benzene ring. Many major contributions
have been made to physical science by men in their twenties and thirties,
whereas contributors to social theory average at least twenty years older.
I interpret this to mean that the frame of reference in physical science
is easy to communicate and internalize, so that one can begin life as a
theorist; in social science, it takes about twenty years more experience
and maturity to develop the background for theorizing.

During this long training period, one finds his goals and develops
a personally meaningful frame of reference. This frame of reference is
made explicit to oneself as a series of propositions about human behavior
— a set of agreements with oneself to look at human behavior from
some defined point of view. Such agreements can carry conviction for

Work-emotionality Theory of the Group as Organism 569

a person only if they come from Ms experience of working with and
puzzling over the phenomena he is studying. And I believe that, unlike
theory building, the development of proposition proceeds most effectively
through interpersonal stimulation, speculation, and challenge.

The problem of characterizing where the field of group study now
stands in its development is complicated by the fact that no two re-
searchers are at the same point. At the present time, it might be valuable
to take each other's propositions seriously. It is much easier to work to-
gether at this initial stage than later, when researchers consider that they
are well on the way to formal systems and therefore have something to
feel defensive about. As I read the nicely organized systems of hypotheses,
derivations, and corollaries so far published, my major feeling is that,
regardless of the amount of supporting research, I wish I knew more
about the origins of those systems. With more metatheory — the proposi-
tions which summarize the researcher's preconvictions — we would prob-
ably find each other's sets of propositions rather similar, even though ex-
pressed in quite different vocabularies. Given such reassurance, each
would investigate in his own way but with expectation that the resulting
hypotheses could be fitted together.

It is the propositions whose fruitfulness has been tested in theory
building that ultimately become incorporated in man's cultural view of
the world, even though it is the theories themselves that have most im-
mediate value for problem solving and social invention.

With regard to the particular research to be discussed in the re-
mainder of this paper, I should say that a series of propositions has de-
fined our approach and has led to a method of analyzing group proc-
esses. As this method is applied to experimental situations, a constant
stream of questions is raised, and these are tentatively answered in specific
hypotheses. Further data are considered in an effort to test the hypotheses.
Generalizations emerge as summaries from many such experiences, and
on the strength of these, the propositions are worked over. Through this
method, the set of propositions is gradually transformed into a body of
theory of rather comprehensive scope (since the propositions were quite
general to start with). Thus we seem to be reaching for a general theory
of human interaction rather than for a special theory of "group be-
havior." The experimental situations focus on groups, however, because,
as mentioned earlier, it is in groups that one sees an acting out of the re-
lationships between personality and societal factors.

STRUCTURE OF THE SYSTEM AS THUS FAR DEVELOPED

The nature of the groups studied. However generalized a system may
ultimately become, it begins in the need of the researcher to comprehend

570 HERBERT A. THELEN

particular phenomena with which he is involved; hence the basic terms
of the system are necessitated by these phenomena. Although we believe
that the system to be described and analyzed is applicable to the whole
gamut of face-to-face operations of all sorts of groups, we must admit
that the reasons for developing the system along present lines become
clearer when one can visualize the sorts of groups we were most con-
cerned with.

Ours have been "human relations training groups.33 These are groups
of 12 to 25 adults who have come together to learn "what goes on in
group" and "how to participate more effectively in groups.55 The
method of training in these groups has been hammered out in a variety
of two- and three-week workshops conducted each summer by at least a
dozen organizations in the United States and Europe. The basic ideas
of training were first put into practice at the National Training Labora-
tory, held at Bethel, Maine, in 1947; and most of the other workshops
have been conducted by staff members who have "been to Bethel." Al-
though no two "trainers" would portray their methods of training in ex-
actly the same language, there has been basic agreement on the funda-
mental basis of operation: that the group studies its own problems of
operation; and that these problems arise from the stresses produced by
lack of structure or definition of the situation in which the group finds
itself. Some trainers produce this lack of structure by taking an almost
completely passive role. Others may alternate withdrawal during some
periods with strong leadership at other times. Still others follow certain
criteria for deciding when and how to intervene.

These groups are designedly heterogeneous with respect to occupa-
tion; "shop talk33 is impossible. The one thing the members have in com-
mon is an interest in group operation. There are no demands imposed
on the group in the sense of particular problems they are to solve; they
must produce their own agenda and deal with it (or not) in whatever
way they can with whatever sort of "help33 the trainer gives them. From
time to time they are led into description and analysis of their own ex-
periences in trying to make or avoid decisions, to deal with nonpar-
ticipants, "problem3' members, or leaders, to plan and carry out acti-
vities, and to study the effects of various individual behaviors on the
course of the meeting.

The groups usually meet for two to three hours at a time, and they
usually hold 10 to 15 sessions during the workshop or "laboratory.33 Fre-
quently there will be an assistant trainer, and the relationships between
the two staff members often contribute to the "dynamics33 (I am tempted
to say "problems33) of the groups. The private motivations of the mem-
bers range all the way from finding out "whether my analysis seventeen
years ago was successful33 and learning "how to be a more effective

Work-emotionality Theory of the Group as Organism 57!

wheeler-and-dealer5 to developing self-insight, skill, and understanding
of group phenomena "50".

Our decision to study these groups was based on the following facts :
since there is great permissiveness, a wide ranee of behaviors can be seen.
The group's effort to understand itself produces much data of the sort
the researcher needs; furthermore, the groups change very markedly over
the course of their meetings.

The development of postulates to research. From our vantage

points of trainer or observer in such groups, we began to develop rather
strongly internalized feelings about what would have to be involved in
the effort to understand them. These "feelings'5 have gradually become
explicit as a set of postulates, and they constitute the metatheory of the
system. The postulates, given in the first chapter of the Stock-Thelen
monograph [44], are

1. Units of interaction exist.

2. Group life can be seen as an adaptive process; and our accounts
of it are descriptions of the changing stresses in the group and of the ways
in which group members respond to these stresses.

3. The emotional aspects of group life, and particularly the use the
group makes of its "feelings" will be direct evidence for diagnosing ten-
sions and the stress conditions which give rise to the tensions.

4. "Individual" factors (e.g., in £ 'personality") will be categories
descriptive of stresses to which individuals are sensitive and of means in-
dividuals use to deal with these stresses.

5. "Group" factors exist by virtue of the interactive networks among
individuals rather than by virtue of what each individual independently
"as a person" brings into the group.

6. The nature of the problems for investigation is given in the in-
teraction between "personality" and "group."

These postulates are both empirical and strategic. In the sense that
they express conclusions from a great deal of firsthand experience, they
are empirical. They are strategic in that they are required bases for study-
ing the phenomena of group interaction. The first postulate is required
to make scientific study possible at the molar level. The second is
required to give structure to the models which we must inevitably use.
The third is a commitment to a psychiatric approach. The fourth and
fifth indicate criteria for a basic distinction to be made throughout
the system. The sixth points toward what we consider the fundamental
nature of the phenomena to be studied, or toward the kind of study
which we think will throw light on all the preceding factors so far
implied.

These postulates are without "content." They are simply agreements
with ourselves as to how we are going to approach the study of groups.

572 HERBERT A. THELEN

They may be compared with the other approaches mentioned earlier
under Current approaches.

Propositions about the nature of phenomena being studied. Over the

years, research guided by these postulates has led to an organized body
of propositions from which hypotheses could be generated. The continual
development and modification of these propositions has been a major
goal of our work. The more specific theoretical concepts and researches
will, we think, be most intelligible when seen against the background
of these propositions.

Since we think that the study of "groups" is only a selected aspect
of the study of human behavior in general, we first offer a set of proposi-
tions about the latter. This then provides a frame of reference within which
the propositions about groups can be formulated. The following discussion
of our propositions is quoted from the most recent summary [52] :

Propositions about human behavior in general

1. Man is always trying to live beyond his means. Life is a sequence
of reactions to stress: man is continually meeting situations with which
he cannot quite cope.2

3 Proposition 1 conceives behavior as purposive. If purpose be assigned to man
as actor then there is required further the concept of something acted on, e.g.,
environment. From this distinction of inner-outer flows also the possibility of the
self-concept as distinguished from the object- or other-concept; and also the ac-
ceptance of a priori realities apart from man. A rigid phenomenological view is
thus inadequate, as is also a strictly interactionist view if the interacting entities
are alleged to be similar in kind.

We see the group as a whole as a system surrounded by an environment and
containing individual subsystems. "Personality" is the term for the unique pattern-
ing of drives or predispositions of an individual subsystem; "tension" is the term
for the tonus or state of mobilization of drive-pattern. The group as a whole is a
"social system" which exerts control over interpersonal and person-environment
(or, more precisely, Dewey's internal-objective) interactions. This control is
exerted through the group "culture," consisting of agreements, perceptual biases,
values, threats, etc., which are imputed or ascribed by the individuals to the
"group" (as superego) for the sake of maintaining order to the extent required
for meeting individual needs and accomplishing convening purposes.

These public or task purposes are achieved through attack on problems to be
"solved," that is, through taking action to change particular conditions perceived
as lying "outside" the group. To bring about these changes, the group must define
and accept two kinds of reality demands: (a) demands for a particular character
of action dictated by the "logic" of the problem and directed against the condi-
tions to be changed; and (&) demands for reorganization of the culture so that the
necessary participant roles can be developed and the needed human resources
mobilized. This latter problem is complicated by the existence of many internal-
objective relationships (such as loyalty to one's ethnic, class, family, or institutional
groups) which are to be maintained while changing the particular internal-ob-
jective relationships whose unsatisfactoriness led to identification of the problem

Work-emotionality Theory of the Group as Organism 573

2. In stress situations, energy is mobilized and a state of tension is
produced.

3. The state of tension tends to be disturbing, and man seeks to re-
duce the tension.

4. He has direct impulses to take action, and there appear to be a
limited number of kinds of impulses "or drive-actualizations). Bion pro-
posed four major purposes or needs of groups and societies to which im-
pulses contributed, and he labeled these fight, flight, dependency, and
pairing. We have since found that the same categories can be used to
describe tendencies within the personalities of individuals.3

5. Impulses may be translated directly into action. This may reduce
the tension and render a person temporarily incapable of further reaction
to the initiating stress. If the stress has objective basis in "real" danger,
then the person remains in danger, and the behavior is nonadaptive. If
the stress is projected from the subjective domain (such as a threat to the
self-concept), then the emotional discharge may be a prelude to reflec-
tion; and the behavior, although not itself adaptive, may make adaptive
sequelae possible.

6. Direct acting out of impulses has varying consequences, depending
on the nature of the impulse. Pairing increases adequacy to cope, with-
out reducing objective dangers. Dependency neither increases nor de-
creases adequacy nor removes the danger; its effectiveness depends upon
whether the sought protection is forthcoming. If successful, fight destroys
the danger, but it also tears up the lawn and makes enemies out of mid-
dle-class persons. Flight gets one out of danger without increasing ade-
quacy or removing stress from the situation. From a long-range point of
view, all these kinds of acting out are mostly nonadaptive because little
or nothing is learned from the acting-out experience.

and purposes. These "hidden" or "process" problems are products of the group as
a system — they come from the social interrelations within the group, and not from
the internal individual subsystem per se.

In our view, the demands of the "hidden" problems, like the demands of the
task problems, result in stresses lying outside the individual subsystem. The group
may or may not have public awareness of a particular stress; different members
may respond in different ways and have different thresholds of sensitivity to a
particular stress. But the "underlying condition," capable of mobilizing each
person's tensions at a given time, is comprehended as a hypothesized stress.

3 Our confidence in the generalizability of these concepts has been increased
through perception of an evolutionary basis for four kinds of impulses. This is
discernible in La Barre, The Human Animal [26]. Fight and flight impulses are
as ancient as the nervous system, and predate the present species of man. De-
pendency and pairing impulses probably developed much later as part of the
psychic equipment for maintenance and reinforcement of familial and societal
(or communal) relationships.

574 HERBERT A. THELEN

7. Impulses may be temporarily thwarted or blocked, and the emer-
gent feelings or other behaviors diagnosed. The nature of the stress is
made known by the behaviors it tends to engender. Instead of acting out,
there is inquiry or "reality seeking." Behavior is mediated by thought
processes in which previous experience is brought to bear, and alterna-
tives formulated, selected, and evaluated in action.

8. All human events contain a blend of acting-out and inquiry modes.
There is acting out in the sense of spontaneous, involuntary expression
of impulse; there is inquiry in the sense of developing awareness of fac-
tors in the situation and in the sense that something is learned from ex-
perience. Bion's concepts of work and emotionality are essentially con-
cepts of inquiry and acting out, respectively, referred to the group as a
whole.4

9. We note that man is capable of using both acting out and inquiry
to meet the stresses of trying to live beyond his means, and wre anticipate
that "successful" adaptation, in both short- and long-range terms, in each
situation requires a particular blending of elements of acting-out and in-
quiry modes.

10. The major theoretical question with respect to human behavior
in general is what conditions tend to predispose men toward modes of
acting out or inquiry; what is involved dynamically in producing and
maintaining an adaptive blend of the two modes?

In our research we are concerned with this question primarily as
applied to the behaviors of groups of people rather than of single ("iso-
lated") individuals or communities. We have further stipulated that we
shall use (initially, at least) the terms suggested by Bion, and that our
method of investigation is to be experimental and observational.

Propositions about the "group." In general our predilections have
been toward an interactive or dynamic approach. Behavior is not literally
a response; it is an event which arises out of a complex system of part-
whole relationships. By "personality" we mean the tendencies for the
individual to be involved in certain kinds or qualities of events. Psycho-
logically, at least, "individuals" are the loci or centers of strains within
the total system. The relief of strain within one part of the system tends
to cause strains in other parts, and this communication or transmission
of strain is mutually influenced by properties of the system as a whole.
According to Bion, the most significant property of the group as a whole
is its "basic assumption of group purpose or need." For about this basic

4 Bion suggests that some amount of work is always present, but that emotion-
ality may or may not be present. Bion is dealing with the culture of a group
rather than with the behaviors of an individual, with molar rather than with
microscopic episodes. Individual affect is expressed even though the culture of the
group may be work-oriented.

Work-emotionality Theory of the Group as Organism 575

assumption the group organizes its expectations, standards, and roles
;Le.? Its culture , As researchers, then, we are concerned with ' 'a - the
group-relevant tendencies of individuals, rb] the way these tendencies
produce a matrix of forces in the group, ' c] the dynamics through which
the "basic assumptions" ernenre from the forces and shift from time to
time, (d] the characteristics of the control system developed to imple-
ment the basic assumptions. Substantively, we are especially interested
in these matters as they relate to the central theoretical problem of deal-
ing with simultaneous capacities and tendencies toward "acting out" and
"inquiry" with each change of the stresses internal to and imposed on the
system.

1. Each person has the capability for meeting stress by "acting out55
and by "inquiry.53 The capabilities differ from person to person.

2. Which particular capabilities or tendencies will be actualized in
the behavior of a particular person depends in part upon the situation in
which he finds himself. There is, however, enough consistency in his
behavior from situation to situation that he is recognizable as the same
personality.

3. Persons come together in the expectancy of mutual benefit in
coping with objective problems and meeting their personal needs.

4. When persons get together, tensions are mobilized5 and interaction
results. Out of the interaction emerge mutual identifications which de-
termine the characteristics of "groupness," including a social order and
structure.

5. The social order exerts control over the interactions among in-
dividuals and gives the interactions a discernible pattern and sequence;
this in turn can be comprehended as necessitated by the group as a whole.

6. This pattern and sequence change in character from time to time,
thus creating the appearance of different units or phases of interaction.
The organizing principle for interpretation of each phase is that the
group culture has shifted distinctively to a different configuration of
"basic assumptions."

7. The culture-units differ in the quality of their blend of "acting
out" and "inquiry55 ; hence they differ in the nature of their contribution
to the group's adaptation to the "inner" and "outer" stresses which were
present initially and which are created as the members live together.

8. The intensity of stress developed in each situation during the re-
lease of tension depends upon the extent to which the mode of tension
release is "acting out." "Acting out" tends in itself to be nonadaptive,
but it builds stress; "inquiry" tends to be adaptive, but it reduces ten-
sion with the minimum development of stress. The problem, of the group

5 Consider, for example, the fact that rather clear soclometric differentiations
are made during the first few minutes of a meeting of strangers.

576 HERBERT A. THELEN

as a whole is to maintain the * 'appropriate5' blend, balance, or oscillation
between these two modes of behavior.

9. As the group continues to meet, individuals adapt to the group
and they adapt the group (culture and basic assumptions) to the indi-
viduals. Thus changes occur in the modal tendencies of the units of in-
teraction. The amount of change depends primarily on the extent to
which inquiry is the dominant mode, for inquiry is associated with learn-
ing. The amount of "group growth" is primarily determined by the
amount of energy flowing into inquiry components of adaptive process.6

10. In general, the potentialities for amount and adaptiveness of
cultural development, and the range of "problems" (stresses) with which
it can deal, are limited ultimately by the "composition" of the group. The
extent to which and the rate with which the group actualizes these po-
tentialities depends upon its "leadership," i.e., its development of means
for controlling and selecting and actualizing needed contributions. In
view of the basic theoretical problem, optimum leadership would strike a
balance between encouragement and support of direct expressions of
affect (so that the existence of stresses could be known) and diagnosis
and bringing into awareness (through problem redefinition) of the fac-
tors giving rise to the stresses to which the group was reacting.

The research tasks. The propositions listed above developed along
with the research investigations; the formulation presented here actually
represents reflection on a great deal of the experience to date. We shall
now move to a somewhat lower level of abstraction and with the re-
search operations and strategy of investigation through which the theo-
retical concepts have been gradually developed and clarified [47, 44, 54].

The first research task was to devise some way to record "what goes
on in a group." We wanted a scheme which would enable us to see what
each individual had contributed, what "phases" or units of interaction
the group went through, with what kinds of tasks, explicit and "hidden,"
the group seemed to be concerned. We wanted a method which would
enable us to deal in the same terms with interchanges between two in-

6 These statements hold best when inquiry is thought of as a conscious process;
for then change would certainly be accompanied by learning and adaptation. The
statements are more tentative in cases where there seems to be change but little
or no learning. One group, for example, developed a culture in which there was
considerable freedom to "fight" but no freedom to "work" — the amount of con-
scious inquiry was practically zero. The group fought for 15 meetings, and ap-
parently never resolved any of its problems. At the same time, however, there were
changes in the way it fought and in its perception that it was fighting, e.g., the un-
spontaneous planning and dogged engagement in "social" activities whose purpose
seemed to be denial of their hostilities. A precise statement would probably be
that the amount of adaptive change is related to the amount of inquiry but that
nonadaptive changes can occur without inquiry.

Work-emotionality Theory of the Group as Organism 577

dividuals, interaction within a gronp over the period of a phase or meet-
Ing, and the whole life of the group.

The second task was to devise some way to relate individual person-
ality to group operation. We wanted a scheme that would enable us to
predict what role individuals would play, and what effects different com-
binations of personalities ;'' composition ; would have on the operation of
the group.

The third task was to devise some way of identifying subgroups within
the total group structure. We wanted to know how far group operation
can be understood as the interaction between subgroups and what part
subgroups play in determining the nature of group operation on the one
hand and of individual participation on the other.

The fourth task was to demonstrate relationships between group
operation and "productivity" or results. We were interested in such
matters as who learns the most from training groups in human relations,
and what sorts of groups reach the "best" solutions to problems.

The sequential method for analysis of group operation. The starting
point was the observation that groups pass through different phases. This
one can feel in his \iscera. At one time the group is tense, confused,
easily frustrated; at another time the goup is happy, relaxed, creative.
These are differences in moods, and they represent different emotion-
alized states of being. It was clear, then, that we needed data on the
emotional or affective aspects of behaviors. In addition, however, these
phases differed, it appeared, in the kind of work that was going on.
At some times, it seemed that every man was intent on his own inquiry;
at other times there appeared to be a genuinely cooperative effort, in
which a whole structure of ideas and conclusions was being erected in an
orderly way. Thus it also seemed clear that we must get data on the
kind of work or thinking that was going on.

In order to reproduce the meetings we wanted also to record all in-
dividual contributions in sequence and along a time axis. And, of course,
we wanted a sound record to fill in the content of ideas that were ex-
pressed.

For a variety of reasons which will be discussed later, we wanted to
get our data on emotion and work by direct observation — meaning inter-
action of trained observers with the live situation — rather than from the
sound record.

About this time, Bion's articles, to which I have already referred, be-
gan to appear, and it seemed to us that he was putting into words the
basic concepts we needed. Our job was to see if we could give these con-
cepts operational definitions so that they could be used for categories of
observation, and a great deal of effort went into this. Bion's basic ideas
were that the mode of operation of a group changes from time to time,

578 HERBERT A. THELEN

i.e., it passes through distinguishable phases, and that each phase repre-
sents a particular combination of basic tendencies to "work" and toward
"emotionality.33 Work is sophisticated reality seeking, using conscious
problem-solving methods, and the ability to work is learned. "Emotion-
ality" is "primitive" unlearned direct reaction, and responds to uncon-
scious needs of the group to maintain itself in the face of stress from with-
in or from without.

Bion [5] suggested three basic categories of emotionality : fight-flight,
dependency, and pairing; these he saw as "basic assumptions" dominat-
ing periods of interaction. We decided to use these categories for specific
individual behaviors as well. We separated fight and flight into two cate-
gories because they are distinguishable individual behaviors. Fight is any
expression of aggression toward the problem, the group, an outside
agency, the leader, the self, or anything else. Flight is any behavior of
"running away" from stress by such means as joking, breaking up the
meeting, daydreaming, incoherent rambling discussion, "academic" pres-
entation, etc. Dependency is shown in any behavior which seeks aid
from outside the person : from the leader, the minutes of the last meeting,
traditions, experts, and so on. Usually, such seeking of aid is accompanied
by expressions of weakness and inadequacy: "the job is too big," "we
don't have the resources," etc. Pairing behavior includes intimate re-
marks made privately to another individual, "reaching out" to others
with expression of warmth, approval, or agreement, or even, by extension,
a warm statement to the group.

We divided Bion's concept of work into four distinguishable kinds:
individual work, in which the person is intent on his own private interest
and concerns; group "housekeeping," in which the group is making rou-
tine decisions about what topic to discuss, how to appoint a chairman,
how long to meet, etc. ; task- or goal-directed work, in which the group
has a clear purpose and is trying to collect information, make suggestions,
select a course of action, evaluate the suggestions, and so forth; and fi-
nally, "integrative" work, which pulls the whole enterprise together. In
integrative work we see the effort to relate what the group is doing to the
"kind" of group it is and wants to be; the statements are thoughtful and
insightful interpretations which tend to give the group its "bearings."
(Looking back to preceding discussion under Common elements we may
point out that the first and second kinds of work refer to the internal
system, the third kind to the external system, and the fourth kind to the
relationships between internal and external systems. )

Thus we have four categories of emotionality and four of work.
After considerable practice, we finally defined the categories by trying to
characterize the intentions and the behaviors typical of each category.
Then we added a good many illustrations of the kinds of behaviors we

Work-emotionality Theory of the Group as Organism 579

meant. When observers were trained to use the categories, it was found
that each tends to have a consistent bias — one sees more fight, another
more dependency, and so on. Since there is no "right" amount of fight or
dependency, the bias simply meant that we should keep the same ob-
servers with the same group during all its meetings. We finally decided to
have two observers rate the behaviors independently, and then, later on,
compare notes and argue out their disagreements, using the sound record
to help them recall the situation.

During a meeting, each observer records the code number of the
participant, the kind of work contribution (one of four categories), and
the kind of emotionality (none, or one of four categories). He also jots
down a few words so that the participant's comment can be found on the
sound record. And every minute he draws a line so that the interaction
is chopped up into equal-sized periods.

The data are then graphed. After much experimentation, we finally
decided to show four broken lines, with a point on each line at the end
of each minute. One line shows how many speakers contributed during
each minute. This we called rate of interaction. Another line showed
"how much" emotionality was expressed each minute — or, more accu-
rately, how many times the observers detected one of the kinds of
emotionality during each minute. We decided, quite arbitrarily, to
distinguish between "big" and "little" emotionality, and to weight these
2 and 1 respectively. Big emotionality is a direct expression; e.g., "Beat
it !" "Little" emotionality is a muted expression which the speaker might
even deny, e.g. "I wonder if Joe wouldn't feel happier somewhere else."
The amount of emotionality each minute was the sum of the weighted
ratings made by the observer each minute. Not all statements were per-
ceived as having "emotionality." The categories of emotionality were
entered in the margin of the graph; only the "total amount" was
plotted.

The third line represented work during each minute. The four kinds
of work were weighted with numbers from one to four. This again was
an arbitrary decision and it reflects the feeling that these four kinds of
work represent a continuum, from "individually oriented" to "inte-
grative." The continuum is one of "maturity" or "group cohesivencss."
During the period of 10 meetings, for example, we find a lower average
work level during the first meetings than during the last; and the con-
cepts of work were first suggested as "developmental levels."

The fourth line is put in for easy interpretation of the work line.
This line represents level two work (group "housekeeping") as a kind
of mundane standard. It is obtained by multiplying the number of
participants each minute by two. The categories of work are also entered
in the margin., and, in addition, the code numbers of the participants.

580 HERBERT A. THELEN

Having figured out a way to plot the data, we became fascinated
with our graphs, and we tried to see if we could detect different phases
just by inspection of the graphs. We found that to some extent we could.
Thus a period of 10 minutes might be found in which the lines were all
close together, and another in which they were spread apart. The closely
set lines would indicate rather equal rates of emotionality, work, and
interaction per minute, i.e., individually oriented work which was highly
charged emotionally. If all the lines were close to the bottom, it would
suggest tension and inhibition; if they were all fairly high up it would
mean excitement. When the lines were wide apart, it would tend to
mean a high work orientation accompanied by low emotionality; but if
the emotionality line was lying on the "floor," the high work might be
pretty academic and uninvolved.

At any rate, we found the graphs, with their rapid overview of the
meeting, quite useful in taking the next step, the observers' effort to
interpret what was going on during each minute. These interpretations
were, in effect, the specific hypotheses which later on would have to be
checked from other data such as the sound record, questionnaires to the
participants, or interviews.

We still needed to find some way to divide the interaction sequence
into phases. If these phases existed and if, as Bion suggested, they
represented different states of being or operation on different "basic as-
sumptions," then it was essential that we be able to identify them so
they could be compared among themselves. A very elaborate analytical
job was done on one series of 10 meetings of one group. We considered
the minute-by-minute interpretations, the appearance of the graphs, the
interviews with the leader and his assistant after each meeting, our own
reactions as observers — and we divided the sequence over the 10 meet-
ings into 120 periods or phases. The whole process was repeated a year
later, when memories were colder, and the differences between the two
"unitizings" were thrashed out into one "best" picture. In the mean-
while, Ben-Zeev [2], a member of the team, worked out an extremely
clever scheme for unitizing the meetings simply from the code numbers
of the participants. He started from the assumption that a somewhat dif-
ferent set of people would contribute during each phase, and he set out
to tabulate the participants and identify the points at which the pattern
noticeably changed. The procedure, which is too complicated to permit
summarizing here, is completely described in [54]. The unitizing by Ben-
Zeev's method was compared with the highly complex unitizing obtained
by analysis, and 74 per cent of his units corresponded exactly with those
obtained by analysis. Allowing a leeway of two minutes at the beginning
and end of each unit, the correspondence climbed to 95 per cent.

Now that we had a systematic and objective procedure for unitizing

Work-emotionality Theory of the Group as Organism 581

the sequences of interaction, it was possible to take the next step, plotting
the units on an emotion-work field. For each unit, we calculated the
average work and emotionality per statement and then located it on a
graph of emotion vs. work. Instead of entering the unit as a point, we
entered it as a rectangle whose center represented the averages; the
length was made proportional to the number of minutes duration of the
unit, and the breadth was made proportional to the average number of
participants per minute. Thus the area of the rectangle indicated the
total number of contributions during the period. Each rectangle was
numbered in the time sequence, and each was connected by a straight
line to the preceding and subsequent units. Thus we were able to see
how the group shifted in its work and emotionality orientations during
an entire 2-hr meeting; we could form some notion of the relative stabil-
ity of each unit, locate transitional units, and note which subgroups of
participants tended to operate in different parts of the work-emotionality
field.

Thus ended the first task : the development of a procedure by which
we could systematically move from individual contributions, to phases,
to meetings — viewing all with the same set of concepts and noting how
one led to another.

The measurement of valence and individual-group relationships. The
notion that different phases of group operation would have different
participants suggests that the phases originate in some kind of common
tendency shared within the "combination" of participants. During a
"fight" period, for example, we might find five people most actively
carrying the ball; during a "dependency" period, four others might be
the active people. One might reasonably think of the five participants in
fight as "fighters" — in the particular situation of the group at that time.

With his psychoanalytic concern over group formation and main-
tenance through object-tie identification, Bion [8] put the matter a little
more complexly, suggesting that certain individuals tend to* "combine"
with each other to maintain each particular mode or pattern of modes.
This tendency he called their "valency." Thus in a particular group with
the other particular members, a person might have a high valency for
fight; i.e., in some way he would "combine" with certain others in the
group to shift the group into operation on the assumption that their pur-
pose at that point was to fight.

The concept of "combination" seemed to us rather difficult to pene-
trate, and we decided to begin by simply trying to work out a procedure
for estimating tendencies toward fight, pairing, dependency, flight, and
work in the group. These tendencies we thought of as an oversimplified
kind of "valency," and we referred to them as the "group-relevant as-
pects of personality."

582 HERBERT A. THELEN

These tendencies are what each individual contributes to the dynam-
ics of the group, and as a result of a good deal of experimentation we
developed a test for assessing them. The test presents the subject with
incomplete sentences each of which he is to complete as rapidly as pos-
sible so that he cannot actually think about what he is writing: we want
him to follow his impulses. The most useful sentence stubs were found
to portray some situation commonly experienced in the group. "When

George attacked the group, Bob " "When the leader offered to

help him, Pete " "Since Jack liked some members more than

others, he " "When Jerry was joking, the group "

"When Sam said 'Let's get to the problem,' I " The stubs also

incorporated the various emotionalities and work. Thus the examples
given are keyed respectively to fight, dependency, pairing, flight, and
work. The various items were chosen to sample each of these modalities
in a range of instances.

The test is scored "quantitatively." In addition, it is capable of con-
siderable "qualitative35 penetration. To compare compositions of two
different groups, the quantitative scores are sufficient; but for prediction
of the behavior of one individual,, the protocol must be studied at some
"depth."

The most obvious first score is simply a count of the number of each
kind of item the person completes with the same modality it suggests;
e.g., how many of the "fight" items are accepted? Of the six fight items
a person may accept all; of the dependency items, he may accept three
and not accept the other three, and so on. By "nonaccept" we mean
that the person completes the sentence with a different modality than
that given in the stimulus. "When George attacked the group, Bob
attacked George" is a clear acceptance of the fight stimulus. "When
George attacked the group, Bob fell silent" suggests psychological with-
drawal, or flight. A second score, therefore, is the count of the number
of times the person introduces each of the modalities into the sentences.

Now just as we found the distinction between "little" and "big" ex-
pressions of emotionality useful in the group, so we found a similar dis-
tinction useful with this test. We often have the feeling that the "little"
expressions on the test are probably unconscious, and might even be
denied. Thus some fighting directed against the leader seems to be tinged
with dependency — the person is fighting the leader as a way to hide
from himself his own feeling of dependency; but he would probably
refer to himself as an "independent" sort of person and would resent the
interpretation just given. At any rate, such overtones could be detected
frequently, and we counted them, too.

In addition, it is interesting to note the form of the response. We
noted (and counted) responses in terms of action, feeling, and ideation.

Work-emotionality Theory of the Group as Organism 583

as illustrated respectively in the following completions of the item "Since

Jack liked some members more than others, he " : "invited them

to have a drink with him," "felt supported," and "wondered whether it
was all right."

In the qualitative analysis, we can go beyond the scores to make
shrewd guesses about the conditions under which a person will have
each of these impulses, the extent of his ambivalence toward or accept-
ance of the impulse, the way he is likely to express it, and with luck,
something about his mechanism of impulse control.

As a result of our investigations with this instrument we have noted
that even our simplified concept of valency contained concepts of three
distinguishable tendencies: (a) tendency to express the modality in the
group (e.g., a person with a "fight" tendency actually fights in the
group), (6) tendency to become actively participant when the modality
is established in the group (e.g., the group is in a "fight" phase and the
person participates, not to express fight himself, but to flee, pair, or
actively "respond" in some other way), (c) tendency to become dis-
turbed or anxious, whether participant or not, when the modality is
established in the group (e.g., the group is in a "fight" phase and the
person becomes immobilized and anxious; in this case he may respond
with several modalities or he may be nonparticipant) .

The sentence completion protocols, inspected by a trained nervous
system, can lead to quite accurate predictions about how most people
will participate in groups. Further refinements of the test are in the
direction of "building in" more situational dimensions so as to increase
the load of interpretation possible from "quantitative" analysis.

The identification of functional subgroups. We found that the hidden
problems of group process— and these are the ones most related to emo-
tional phenomena — could be viewed as representing a struggle to estab-
lish the work-emotionality assumptions on which the group would oper-
ate. But there were two problems that we felt needed investigation. As
diagnosticians, we needed to conceptualize the struggle, not only in terms
of emotional tendencies in group culture, but also in terms of much more
precisely defined purposes. One wants to know not only that there is a
struggle between the fighters and nonfighters but also what the struggle
is about; e.g., those who want a "strong leader" vs. those who would
like to be leaders themselves, know that they cannot, and therefore
compromise by trying to prevent anyone else from becoming "leader."
Such concepts of purpose would enable us to bridge between individual
emotional tendencies on the one hand and the content of discussion on
the other.

The second problem was the unsolved part of Bion's notion of va-
lency, namely the idea that persons "combine" to maintain the balance

584 HERBERT A. THELEN

of work and emotionality characteristic of each phase. What is the
basis of this combination? Although it might be ties to the leader, this
would not account for differences in participation patterns. The notion
that the participant subgroups represent sociometric cliques was found
to be true under certain conditions, e.g., when the basic assumption is
pairing, and untrue for others, e.g., when the basic assumption is fight.
The notion that there is identification with some "central person" [37] or
"spokesman" [6, 9] seemed reasonable, but we found the concept diffi-
cult to use in any confident or systematic way.

We decided to investigate the basis of formation of the shifting sub-
groups of participants during the various phases. The first problem was
to find some way to represent individual purpose at a level less "deep"
and nonspecific than emotional tendencies but more "deep" and funda-
mental than consistencies in overt behavior. In terms of "levels" or
"depths" we needed something analogous to the TAT whose usual in-
terpretation lies between the deeper Rorschach on the one hand and
what the man says on the other.

For this purpose we adopted the self -perceptual Q sort [42]. The
individual is handed a pack of cards each of which contains one pos-
sible self-perception; e.g., "I tend to dominate when the group gets con-
fused." Each item was keyed to the more fundamental modalities, so
that ultimately the data could be related to the deeper need system. The
individual sorts the pack into a specified number of piles, with a specified
number of cards in each pile. Thus he is forced to distribute the items
within a normal distribution curve, which makes the statistics easier. At
one end of the distribution are the cards "most characteristic" of him-
self; at the other end, the "least characteristic" items — as judged by
himself. Through factor analysis it is possible to find subgroups of people
whose distributions show the same common factors, which in this case
are themes or purposes. Through further work, involving analysis of the
contribution of each item to each factor, it is possible to describe with
some accuracy what the common factors characteristic of each "sub-
group" are. With this information, it is then possible to view the events
in the group as the acting out of inter-subgroup dynamics. Conceptually,
not statistically, this is an immense simplification as compared to trying
to account for all of the interpersonal dynamics; and we believe we are
on the track of some highly practical notions. The data obtained through
this method of approach make possible a close rapprochement with the
concepts of group and subgroup culture as understood by the sociologist,
and may thus further more explicit use of sociological concepts in the
system of thought.

Demonstration of relations to fc productivity." We, too, are interested
in the problem of whether certain kinds of group operation and indi-

Work-emotionality Theory of the Group as Organism 585

vidual valency patterns lead to more learning, better problem solving,
and the production of canned peas at a lower cost. Efforts to answer such
questions would invest the whole structure of ideas with social value.
Since evaluative assumptions frequently are smuggled into the study of
groups, we considered it effort well spent to study whether, and to what
extent, evidence actually supported these implied value judgments.

The groups wre have studied were "human relations training groups."
They were selected for study because the normal course of their opera-
tion provided considerable material about feelings, anxieties, ideas of
emotional blocks, and other data often hard to secure. The productivity
of such groups lies in changes within individuals, however; and these
changes in attitude, self-perception, and world view are harder to meas-
ure than, say, educational achievement.

Despite these difficulties, however, there have been sporadic forays
into the realm of the practical. The nature and efficiency of decision
making was compared for two groups, one of "flight" people, the other
of "work-pairing" people (the latter group was "superior"). The work-
emotionality cultures of 50 groups were compared and related to the
quality of their problem solving. (The fight- work group with "integra-
tive control systems" were "superior"). Two training groups were com-
posed alike but "pairing" tendencies were left out of one (it had a
most difficult time). Individual valency patterns were also related to
many fragments of evidence about "trainability." Positive indicators
were acceptance of tendencies toward fight and pairing and existence of
un denied emotional conflict; negative indicators were tendencies toward
dependency and flight and immobilization.

As far as they go, these studies are instructive. They are valid in the
sense that predicted differences in outcomes were found. And they con-
firm our evaluative reactions. But a great many more, developed within
a systematic program of inquiry and demonstration, are needed to estab-
lish the social utility of the system.

From private to public domains: "Blind analysis" and theory con-
struction. As indicated earlier, these studies have been used to refine and
tighten the system of propositions underlying the research. In the course
of the studies many specific hypotheses have been generated and tested,
but so far this experience has been fed back into the propositions rather
than built into a system of theory as usually represented.

This practice is related to the problem of moving from private to
public worlds. It is not enough to make and confirm specific predictions
(hypotheses). One must also show how these specific hypotheses are
derived from general principles, and this he must do in such fashion
that other competent people can do the same. Here lies the major scien-
tific problem of clinical psychology, a problem which is ours as well.

586

HERBERT A. THELEN

How, for instance, does one publicly demonstrate his movement from
Rorschach data to predictions of behavior?

There are two approaches to this kind of problem. One may note
the principles he seems to be using, record them, and then try to organize
the most frequently used principles into a coherent body of main and
subpoints, e.g., general hypotheses and corollaries.

Or one may use a second approach, one we have explored far enough
to consider it worth reporting. This approach assumes that the key to
coherence lies in the relationships within the phenomena and that con-
ceptual relationships should reflect these coherences. For the criterion of
social utility (frequency of use) implied in the first approach, we sub-
stitute the criterion of conceptual necessity. We suggest doing this
through "blind analysis." The proposed operations are as follows:

First draw the graph of a meeting, as explained earlier, (see The
sequential method for analysis . . . ) . Then analyze the "group-relevant
personality tendencies" of each member, using the sentence completion
test administered individually and privately; and also identify the sub-
groups and their purposes (see The identification of functional sub-
groups) . Then hand these sets of data, not including the sound record or
any other record of "content," to another researcher with instructions to
reconstruct the dialogue of the meeting. After he has done this, his re-
construction can be compared with the actual dialogue, and checked,
contribution by contribution, with the record. The instructions to the re-
searchers at all four steps are to "think out loud," preferably in front of
a recording machine. The thinking they must do to accomplish the job
should then be analyzed and organized to produce the body of "theory."

The logic of the approach might be put in this form : if our proposi-
tions which gave birth to the research methods are correct and if our
methods truly reflect these propositions, we may assume that all the
necessary data are on hand. It follows then that thinking is the task still
to do; the body of propositions and concepts (with their large surplus
meanings) gives the general orientation, sequence, and conceptual tools
for the thought required. The product of analyzing the thought process
should be the desired theoretical structure. To meet the argument that
this is too long a chain of thought between firm data, further experi-
ments can test particular, crucial principles in much more limited
settings.

We have not carried this proposal through in all its particulars. We
have, however, very successfully reconstructed, as a "blind" operation,
the dialogue of several short sequences of meetings. We used the graphs
of the meetings but substituted for personality data from the tests, knowl-
edge of the participants gathered from our own observation of them in
meetings. We have been able to expose our thought processes in putting

Work-emotionality Theory of the Group as Organism 587

the two kinds of information together. In other words, the £ 'principles53
to be organized in theory are close at hand. This preliminary check per-
suades us that we have properly identified and understood the required
data — and certainly increases our confidence in the usefulness of our
propositions.

The systematic independent, intervening, and dependent variables.
Inquiry into human behaving proceeds in several stages. First one ob-
serves behavior, these observations are his data. Then one looks for pat-
terns, consistencies, or themes within the behavior, e.g., directions of
growth, disequilibria, projected goals, etc.; these ideas of coherent or-
ganization of behavior are the independent variables, and they sum-
marize tendencies toward change. Then one reflects on these tendencies,
tries to "explain" them and indicate what sorts of things will happen
next. The concepts used for reflection and explanation are intervening
variables; the various sorts of things which might happen next are de-
pendent variables. The researcher selects certain aspects of the possible
next events that interest him and tries to define what he would actually
observe if his guesses were right. And this prediction now is of the actual
data he expects to collect.

All of this is guided by previous thinking, in which the researcher has
asserted probable relationships between the independent and dependent
variables. Such hypotheses are general. From these he could also state
very specific hypotheses, that certain behaviors will be followed by other
equally specific behaviors, but such hypotheses are not useful unless the
particular behaviors selected are clearly symptomatic of organismic pur-
poses; that is, unless such behaviors are meaningful.

In our work, then, we can say two things about our variables:

1. The independent variables are tendencies and therefore imply
further behaviors. The intervening variables are constructions of organis-
mic purpose., usually in terms of uas if." The dependent variables are
classes of behavior that will follow as the organism moves on to achieve
its purposes.

2. The "organism" may be either an individual or a group, in the
sense indicated under Common elements.

Listing the three kinds of variables presents difficulties, in my opinion,
because the classification independent-intervening-dependent arises from
the course of inquiry rather than from the nature of the system. Certainly
the variables called "dependent" in one investigation might be ^inde-
pendent" in another. As explanatory constructs, the intervening vari-
ables, however, could never be anything else. Thus one can start with test
scores on the sentence completion test described earlier and predict be-
havior in the group, but one could start from behavior in the group and
predict test scores. This sort of reversibility would hold in any investiga-

588 HERBERT A. THELEN

tion that tries to predict behavior in one situation from behavior in
another situation, and this is what our researches typically boU down to.
We may limit an investigation to a part of this process, as when we try,
from specific behaviors, to infer the structure of the group ; yet sooner or
later somebody must ask the question: if my picture of the structure is
correct, then what should I now observe? So that we end with behavior,
if only to validate our intermediate larger conceptions.

With this orientation, then, I may analyze the variables in the four
actual research designs described on the preceding pages. We may then
examine in more detail the nature of some of the variables that are
crucial in the development of the system.

The independent variables. The table shows four independent vari-
ables. Two of them are applied to the "live" observed group, and two
to questionnaire or test results, obtained from each individual. One of
the variables from each situation has to do with tendencies toward the
various modalities of emotion and work. These two variables are alike
in concept, as will be seen from the description in the preceding section.
The variable "explicit purposes" is a low-level generalization from the
content of discussion. It comprehends the explicit ideas expressed by
members as to what the group is doing or trying to do. The fourth vari-
able is the set of relationships, obtained by intercorrelation, among the
Q-sort distributions of the members of the group. It implies a whole
series of specific overlappings (to various degrees) of the perceptual
fields or "life spaces" of the members of the group.

The intervening variables. These are more interesting because they
represent key theoretical ideas. They do two things: they collate the
items of the independent variables into wholes or patterns; and they
enable us to shift from one situation to another. That is, they move from
individual properties to group properties, or from individual behavior in
a private situation to individual behavior in a group situation.

Thus "basic assumptions during unit" is a diagnostic generalization
about the e-w pattern of the group culture — the assumptions on which
the group seems to be operating; but it is obtained from generalizations
of the minute-by-minute relationships between collected individual ex-
pressions of work and emotionality. When these basic assumptions,
which imply unconscious "purposes" of the group (mainly in the realm
of self-maintenance) are put together with generalizations about the
explicitly stated purposes of the group, one is enabled to make a fair
diagnosis of problems the group is working on, and the conflicts and
strains within the group as a whole. These "intervening variables" then,
stand for the application of a good many principles from individual and
group psychology; the diagnostic mind is not a machine for automati-
cally processing the input of independent variables. For rigorous system-

fl

, M

rt

a

1

§ .5 fl A "5

.s

fl

"fl

M Q Q M-( ^|

^3

'rt

.§

'.§

8

O co "ti O TJ

fl

^

.§

"3
c
fl

*o

§

§^1 ^ ^ g

_,

•y

Si

o

""P

0

"*j ^ fl "^ y *rj

jH

]>

fl

"S

c3

^ ^H TJ t-< <tf

fl

1

rt

1

'•g

1
o

0

a

1

S
)-H

l-^l- o-2

ft

i

fl

Q

o
fl

^

^

1

S

CX

a

1

OH

2

bJD

structur

of meet-

ft

4

1S.IP

! iff Ji

to

a

Predicted e

trends
Predicted g

poses
e-w role in

Sociometric

§

T!

(U

P-*

ing, others
Identified si
members

"themes"

i

o

'i ii 1

vening

-<->

o

D

S-i

i

•£

ft 4J

8 'fl

.2

T3

common

ji* 1*1

£

fl

fl

a

CO ij£
0 *3

1-

fl

a
O

!

Valence

Valence

Factors of
ity

i

TJ

TJ

TJ "d TJ

_a .

S

fl

^

fl fl fl

fl "0

<u

<u

aj

5-1

o aJ o

o

.2 g £

-S CQ rH

td . 3

1

TJ
cd

|

TJ Tl TJ

'ft £ S-4

cd cd cd

fl

fl
o
U

i
Shifting rel
between e
trends in i

a
.'§

ii

x

to
TJ
fl

o

Trends tow
w. modes

Intercorrelc
tionships

? CO ?* C/3 ? CO

0 <U 0 flJ O S
•*-* TJ •M TJ •*•* T)
to O co O co O
TJ fl T3 fl T3 fl
fl q fl C fl q
D • flj * O •

ft

„!

•t-j

4J +J j_)

CJ

§5

CO CO CO

TJ

^

S3

&

OH

+- +J &

hH

1

^

c3

d

53

d o o d

^

OS
03

BJ

rcj

r<

M

fl
o

8

(U co

CO CO "^ CO

O

Q

: Observer's
and w. *

i

1 Verbal bel

Responses

I
! Responses

CO rj

*o £

tl I!

o c3

§j OH qj (U <3 P

S"° PH g § ^O
pd Q£j C^j

TJ

fl

&

i

o 8

.2

^

t)

fl o

P

3

CO

2

I-J

fl

.S

2

PK

OJ5

TJ

* « +J

'•C

"3

[T|Q

O

£ fl

c

.£

3
0

§

g

^ft

T3 c^

"•j-

c

3

*c3

13

L

9 'cj

J

P

H

3

•43
1

8
§

fl
.2

o

1=

O co
* -1—

tr

fl

<U .tl

CO

!>

fe

PH

589

590 HERBERT A. THELEN

atization a great deal of breakdown of ideas is required, as suggested in
design 5. And tearing ideas out of their gestalts does not impress me as
easy; it may even be perilous.

The intervening variable of "valence," which has been discussed be-
fore, moves from private behavior to public behavior expected in the
group. Our concept of "valence" tends to operate with two degrees of
complexity: through rather low-level interpretation, one can tell from
the "quantitative" analysis a good deal about the circumstances of prob-
able participation as well as the e-w nature of the participation — for
most people tested. When one adds the "qualitative" analysis, the aim
is to penetrate the threat-defense system of the individual and to make
richer inferences about his relationships or identifications with others in
the group. It was more in this latter sense that Bion suggested the term,
but the former, simpler sense has been adequate for some purposes.

To my knowledge an outstanding property of valence is not shared
with any other variable thus far suggested. Valence seems to be the one
variable whose individual measures can be averaged arithmetically to
get "basic assumptions" of the group — and this is a group tendency.
This finding is extremely useful because it enables us to compose groups
for various purposes almost at will, an important practical achievement.
This property derives from the fact that individual emotional tendencies
directly influence — in fact, produce — group emotional tendencies. Never-
theless a more adequate explanation must wait until we know more
about the way affect is "shared" and about the nature of intermember
identification.

The factors of "commonality" refer to the third design described
earlier. They enable us to move from consideration of the individual as
a person to consideration of the individual as a member of a subgroup.

Behind these intervening variables are basic propositions about stress
and tension in human behavior. As referred to members, the intervening
variables describe tension within the member; as referred to the group,
they describe stress within the group. For the latter concept Bion uses
the term culture (basic assumptions) which describes the psychological
purpose (whether conscious or unconscious, accepted or denied) of the
group as a whole.

Following Bion, stress arises from capacity for two basically different
types of behavior: emotionality and work. In each situation, the group
is confronted with the problem of working out a relationship between
these two aspects of its functioning. On the one hand, group members
have the need to listen to each other, to bring previous experience to
bear on their problems, to think critically, to diagnose problems and deal
with them explicitly; this is work. On the other hand, the group has the
need to maintain itself as a group: to flee from emotion it can not

Work-emotionality Theory of the Group as Organism 591

handle, to develop interpersonal relationships so that individuals may
deal with their own private anxieties, to find leadership on which it can
depend for maintaining social order, to express its aggressions and re-
sistances in situations with which it cannot cope. Different individuals
have different valencies for these kinds of behavior, and somehow a
balance must be struck among them. This balance, macroscopically de-
scribed as the pattern of "basic assumption" on which the group is oper-
ating, is its "culture" during each period of operation.

Working out the relationships between emotionality and work means
dealing with ambivalence, conflict, and anxieties. In other words, it is
accompanied by stress. Each individual reacts to this stress in his own
way. Thus we may notice that during a certain period, Mary is express-
ing dependency, Joe fight, and Harry is pairing with Tom. These are
individual modes of reaction. We inquire, under what conditions will
each of these individuals behave in this way? We can list (from the
sentence completion test) a number of hypotheses about different condi-
tions which would be likely to evoke these responses. The one condition
common to Mary's dependency, Joe's fight, and Tom's pairing would
seem to be the group's culture. This, in fact, is the safest way to diagnose
the basic assumptions operative in the group. Thus, this concept of basic
assumption is also a concept of a stress or pressure on all the members of
the group, or, if you prefer, a group need to which all the members
react. Yet this concept of group need or stress may not be within the
perceptual field of a member; we note that some members are usually
quite unaware of this sort of influence even though they appear to react
to it.

If we extend the notion of group need to the idea that the need may
be within an internalized group rather than only in a face-to-face actual
group, then we have described something very similar to Le win's "alien
factors" which are outside the perceptual field of the individual even
though they influence his behavior. The notion of a basic assumption
within an internalized group might also be a "press" in Murray's sense.

The individual, with his own predispositions, finds that the stresses
in the group act as triggers or cues for selection of the particular predis-
positions that will be translated into behavior. This need for selection
among his various capabilities and desires puts the group member under
tension. The tension may represent ambivalence, anxiety, fear, anticipa-
tion of reward, and so on. The individual may step in boldly to estab-
lish whatever modus vivendi is most effective for reducing his tension.
He may withdraw from the group or he may adopt some middle ground.
Whatever he does, there will be other individuals in the group who in
whole or in part "want" to adopt the same mode. Subgroups thus may
emerge, one seeming to advocate discharge of tension through fight,

592 HERBERT A. THELEN

others through work, flight, or some particular blend of these. The domi-
nant subgroup is the one most participant at that time; its preferred
mode becomes, in effect, the basic assumption ruling the group's culture.

The matter is further complicated by mechanisms of denial and other
secondary7 dynamics that are themselves productive of further tension
for individuals and of further stress in the group. Thus when fight, for
example, becomes established as the basic assumption, there are those
who at some level would like to fight but who repress this desire — often
with the expenditure of much psychic energy. In general, the person or
subgroup seen as a "problem" by other members is one who expresses
impulses that others are trying to repress.

Dependent variables. These are behaviors elicited during or after the
meeting. During the meeting they are cognitive and affective contribu-
tions in the various categories of work and emotionality. After the meet-
ing they are postmeeting perceptions of self, others, leader, task, group,
critical incidents, periods of success and failure, and so on. These may
also.be elicited in special test situations for the group, such as the picture
projection test designed by Horwitz [22].

The dependent variables shown in the analysis of experimental de-
signs are, as previously noted, classes of behavior in a new situation or
under the new circumstances to which we are trying to predict.

Mode of definition of representative variables of each category.
1. Independent. The independent variables represent general tendencies
for emotionality and work responses to stress situations. They have been
arrived at both empirically and by interpretation of widely varied knowl-
edge. Fight and flight, for example, are primitive states of mobilization
of the nervous system in stress situations. Dependency probably became
internalized with the development of the family. Pairing can be seen
both as a sexual and a societal manifestation in the development of new
families and in the establishment of the sort of societal interdependence
increasingly required for survival of the species. These basic modes, then,
are defined at a high level of abstraction. Operationally they are de-
fined as the "purpose" or "intention" motivating a wide variety of
behaviors.

The definition of the independent variables is based upon one of two
assumptions: (a) behavior is purposive and at a high level of abstrac-
tion there are a limited number of basic purposes of human behavior;
(ft) we cannot understand behavior (because of our own human consti-
tution) until we have named its intent, and the intents we note are those
we are capable of noting because we are concerned with them in our-
selves. In either case, the result is the same; the "definition" is some-
where internalized within the researcher, and through his research ex-
perience, he gradually tries to make it explicit in language. The prob-

Work-emotionality Theory of the Group as Organism 593

lems of definition in our work are essentially similar to those in clinical
personality study.

Any practical means for controlling the valencies in the group uses
empirical independent variables. Thus the "composition" of the group
can be controlled by deliberately selecting people whose individual va-
lency patterns are alike or different in specified respects. Or, the compo-
sition may be built up out of people who will be "effective" together on
a task or in evading a task; here one begins with some one person's
valency pattern, and then considers the sorts of additional patterns that
would be supportive or threatening. Even without specific knowledge of
individual valency patterns, a great deal can be done roughly with class
and ethnic background factors. Over time, we hope through cross-
cultural and class studies to spell out some cultural characteristics in
these terms.

Besides controlling the opportunity of members to "combine" with
others (through choice of those others), further control can be exerted
by setting up institutional expectancies for the group and by defining its
task. We have, for example, worked with "human relations training"
groups because these operate in a climate of the broadest range of ex-
pectations. In a way, anything goes; hence we can get fairly intense
expressions of many tendencies that might otherwise remain suppressed.
Comparisons can be made of a group with wide limits to expression as
against one in which all autistic behaviors are "out of line." This sort of
comparison involves not the selection of the initial tendencies but rather
the control of which among them will be allowed expression and thus
contribute data.

2. Intervening. The intervening variables are concepts of arrange-
ment of tendencies. In a final development, systematization of these
variables might well imply "force field" analysis of the sort Lewin was
driving toward. Currently, the possible interactions between two tend-
encies are considered to be reinforcement, conflict, or repression; in
addition there are reactions to these possibilities: action, anxiety, am-
bivalence, immobilization, and the like. The intervening variables be-
come the key concepts in the theoretical reconstruction of events in the
group. This reconstruction is complete when it explains why each partici-
pant in the situation interprets it the way he does.

With intervening even more than with independent variables, the
researcher's sensitivity and training provide definition. The constructs of
"valence," with its innumerable possible ramifications through "qualita-
tive" analysis, and of group basic assumptions, with their implications
of underlying psychological dynamics, mean more to some researchers
than to others. In other words the constructs have "surplus meanings."
A theory, with carefully defined intervening variable constructs may

594 HERBERT A, THELEN

in the hands of one person "mean" little; in the hands of another, it will
explain much. For a theory- is not a written document, it is an internal-
ized set of tools for dealing with the practical business of meeting certain
needs. This is particularly true in the study of human behavior, which
we must always to some extent see from the "inside" and be personally
involved with.

3. Dependent. The dependent variables lead into predicted specific
behaviors and are measured by counting frequencies in broadly defined
categories. They are defined usually by setting up continua of behaviors,
illustrating at one end an extreme manifestation, at the middle an in-
difference, and at the other end extreme denial or rejection. Such con-
tinua can be set up for aspects of behavior relevant to whatever group
or individual product is being used to supply the performance criterion.
In so far as possible, we select dependent variables which are relevant to
the purposes of the group, for such variables produce many more be-
haviors which can be recorded as data. Purposes may include change of
individual skills, acquisition of knowledge, and change of value-attitudes
of individuals, development of group leadership, group culture, or group
efficiency of operation, and finally, the production of minutes, decisions,
project-activities (such as role-playing), and so on. In all instances the
data are behaviors, but the process of interpretation is guided by the
methods of abstracting and collating described in the monograph men-
tioned earlier [54].

Major interrelations among constructs. At this point, we may at-
tempt a more concise statement of the interrelationships among the con-
structs. From data, through independent variables, to intervening vari-
ables, we are constructing part-whole relationships. From intervening
variables to dependent variables to predicted outcomes, we are construct-
ing whole-part relationships. In this fashion, we move up and then down
the "ladder of abstraction."

Specific behaviors elicited in response to a large number of specific
situations (as in the sentence completion test) can be viewed as parts of
larger tendencies to deal with stress through emotion-work modalities.
These tendencies, as generalizations, collect specifics into a more abstract
whole, a pattern of probabilities; these are represented as "independent
variables." Both the specific behaviors and the tendencies refer to the
same thing: a particular individual acting alone.

As we move to the intervening variable of "valence," we move to
another domain, the group, rather than the individual. The independent
variables, referring to each individual, become parts in a larger, more
abstract whole, the group. The concept of valence includes (nonexplicitly
as yet) the concept of identification through common object ties — which

Work-emotionality Theory of the Group as Organism 595

in this case serves the purpose of maintaining the group through estab-
lishing a particular combination of "basic assumptions" of group pur-
pose. From the larger whole, the organizing principles of the group
culture, the chain of deduction begins. The dependent variables identify
categories of behaviors that will, presumably, be found from the inferred
pattern of basic assumptions. The data at this end of the bridge are
once again specific behaviors, but they are behaviors to be observed in a
particular group composed of given individuals. Thus the over-all move-
ment has been from specific behavior of individuals in private situations
to specific behaviors of individuals in the group, or, from individual
tendencies in "personality" to tendencies of members as part of an inter-
dependent system.

INITIAL EVIDENTIAL GROUNDS FOR ASSUMPTIONS OF SYSTEM

It is clear by now that we tend to consider research as merely one
way for a researcher to meet real and important needs of his own. The
development of a theory may, therefore, reflect quite basic factors within
the researcher's nervous system. In any case, preliminary work and prac-
tice as "trainer" in many groups led to certain "convictions" which had
the force almost of axioms. These notions led to the decision to see what
could be done in experimental situations with Bion's concepts. The "con-
victions" were

1. That groups do have periods in which they are dominated by
different moods.

2. That the concept of group qua group rather than as group qua
collection, although a theoretical construct, is essential to our thinking
and reacting as social beings.

3. That the "laws" governing group life will be "laws" of change:
that is, they will (when developed) be concerned with the continually
shifting balance of forces in the group, and with the continually shifting
"culture" (in the sense of unconscious values and purposes) of the
group.

4. That affective behaviors communicate directly and nonverbally
and are sensed directly, i.e., that "emotion" should be recorded as pri-
mary data rather than used to "explain" so-called objective behavior.

These convictions seemed to summarize our experience with groups.
They served as specifications which the research would have to meet:
(a) for sequential analysis, or analysis of "flow" of experience through
time, ( b ) for finding concepts that would fit the group as a whole — and
by this I do not mean simply analogies to individual personality, (c)
for seeing in the group a dynamic interplay between conflicting tcndcn-

596 HERBERT A. THELEN

cies, I.e., for looking for some dynamic, dramatic theme in group life,
and (d) for categorizing emotion directly through paying attention to
our own visceral responses to goings on in the group.

All four of these convictions pointed toward a "psychiatric" approach
to group dynamics as distinguished from sociological, perceptual, socio-
metric, and other approaches. In Bion's stimulating articles, we found
many of the concepts we needed. Bion's terms fight-flight, pairing, de-
pendency, and work were put forth to describe the moods that groups
sustain at different periods. The notion of applying these terms to char-
acterize specific individual behaviors was not suggested by Bion; he was
primarily concerned with group modalities and individual tendencies.

To us, the concept of the group as a group seemed also to be more
convincingly suggested by Bion than by any other theorist we knew.
Especially useful were Bion's ideas that the individual in some sense is
always reflecting needs of the group, at least during some periods, it is
as if the group were speaking through many voices and the particular
individual whose vocal cords are thus utilized is relatively unimportant
as an individual. Then, too, the notion that at times different people
are "spokesmen39 for the group, although frequently discussed by others
in connection with "what is leadership," seemed in Bion's thinking to be
extended and generalized to cover a great many other possible roles as
well. It also appeared that Bion's concept of unconscious identification
with subgroups attempting to maintain or promote particular basic as-
sumptions (of emotion- work) made room for a subgroup structure which
fitted the facts of group life more adequately than do more sociological
or sociometric concepts (although these clearly are useful at times). But
primarily Bion located an idea of the group as an organism in its emo-
tional "sharing," its continually shifting member identifications, and its
different moods.

Bion does not discuss the concept of the nature of the laws we should
seek, but he does speculate about the problems that such laws would try
to encompass. The question of what brings one mood to an end, and
what initiates another, seen in terms of group anxiety on the one hand
and individual valence on the other, seemed to us to be central. So far
as I know, Bales [1] and his associates, using his "interaction process
recorder" offer the only other way of looking at this problem; but the
failure to identify "natural units" seemed to us to make the system less
useful for our purposes.

The request to "identify the chief classes of experimental and/or
empirical data which have served as the initial source of evidence on
which the system was based, or which have been used in any way to sug-
gest the major assumptions of the system" presents me with difficulties.
A system is not based on data; it is based on more primitive systems.

Work-emotionality Theory of the Group as Organism 597

System building, of course, involves the use of data, but the organiza-
tion of the system lies in meta-theoretical thinking, and it is this which
makes a system a system.

During the period 1946-1950 we carried out a number of experi-
ments which now may be seen as preliminary to the present work. Those
experiments helped us define "major assumptions,55 less by "suggesting35
them than by "bringing them into the open.'5 Thus a whole series of ex-
periments deepened our implicit conviction that data on emotions and
attitudes expressed by group members are central and primary. The
work of Withall [56], Flanders [15], and Perkins [36] showed that
numerous dependent variables can be predicted from knowledge of the
teacher's "intentions55 to support and help the pupils as distinguished
from supporting and helping himself. (Actually, he needs to do both, but
this was a useful initial approach.) Steinzor [41] and Blocksma [10]
also developed categories of "leader intention55 along the same lines.

The notion that purely objective data were of little value was also
tested, primarily because the research problems involved in dealing di-
rectly with emotion as data are troublesome, and we wanted to be sure
that it would be really necessary to tackle all the problems of bias, sub-
jectivity, and so on. Accordingly, Marks [32] made pictures every 15
sec by time-lapse photography, and then counted such overt behaviors
as can be perceived visually. For example, he counted the number of
frames which showed each student standing within 3 ft of the teacher
during a "creative arts55 class. In this class, the expectation is that stu-
dents are very much on their own, and it seemed reasonable to suppose
that a pupil who spends a great deal of time in the vicinity of the
teacher — possibly "waiting for55 the teacher — is probably a rather de-
pendent child in the arts class situation. The purely objective evidence
enabled Marks to pick out several pupils most often found close to the
teacher. Marks then had all the teachers of these same children rank
them in order of their tendency to "be dependent55 on the teacher. The
teachers5 judgments showed practically zero correlation with Marks's
ratings. Moreover, interviews with the students showed that many hy-
potheses other than "dependency55 may explain why a student keeps
close to the teacher.

Such a study is consonant with the belief that, by themselves, com-
pletely objective data are essentially meaningless; hence they must be
complemented with other kinds of data which give some clue as to the
subjective meaning to the actor of his overt behavior.

At this point, we had a choice as to where, in the chain of think-
ing, to introduce the concept of purpose, attitude, or intention. The two
possibilities seemed to be: (a) collect data, however subjective, in the
actual situation— thus introducing the subjective material right at the

598 HERBERT A. THELEN

beginning; (6) regard emotion or purpose as a construct (probably an
intervening variable) to account for the observed objective aspects of
behavior. We rejected the latter because it seemed to us that no new
theory could ever be produced from it.

Thus our empirical independent and dependent variables refer to
"what is going on," but this is understood to mean not only "what is
the overt behavior" but "what does it appear to mean to the actor33 and
"what does it appear to mean to the others in the group.33 Affective
aspects of behavior are rated because these overtones imply both how
the actor evaluates the situation in which he finds himself and also what
his purposes or needs are. In addition, we wanted to have a record of
the activity engaged in, that is, the aspects of behavior related to ex-
ternalized purposes or needs. The categories of emotion and work are
the chief empirical variables, and they have been described earlier in
this paper.

But emotion and work are not completely empirical categories. They
imply much more than a simple, specific, single, univocal, unitary act:
they imply a need, an expectation, a kind of feeling about self, and so
on. To make a rating is an act of comprehension, not simply an act of
recognition. The distinction between "empirical33 and "systematic33 in-
dependent or dependent variables can be made only in the grossest
terms, so that I would be quite puzzled over where one begins and the
other leaves off in our work. I might say that, since our ratings can be
counted and summed and divided by n, all the manipulations that can
be done arithmetically define the empirical portion; the place at which
the researcher's intelligence becomes necessary then defines the "sys-
tematic33 part of the construct. This may accord with the spirit of our
age of specialization, but it is probably an industrial rather than a
methodological argument.

CONSTRUCTION OF FUNCTION FORMS

Generally speaking, the precise mathematical expression of linkages
between constructs is the most elegant way to state hypotheses for pin-
pointed, effective demonstration. This is clearly a sophisticated step
which can only be taken after the system has been developed rather
thoroughly. Our system has not been developed to the point of precision
required for this kind of treatment.

There are two bases for "function forms" within our system as de-
veloped so far. One basis is methodological, the other prepositional.
The methodological basis is a clearly defined sequence of operations that
the researcher carries out; that is, the course of inquiry is rather com-
pletely described as a series of steps each of which summarizes pre-

Work-emotionality Theory of the Group as Organism 599

ceding steps and anticipates next steps. This has been described at some
length earlier. The procedure provides a set of operational stipulations
among constructs; i.e., the constructs are interrelated through what the
researcher does with them rather than through "constitutive definitions"
apart from the researcher. The prepositional basis is a sequence of state-
ments which guide the inquiry conceptually rather than operationally.

Our stipulations with respect to the interrelationships among vari-
ables thus tend to be "general adumbrations of the functional relation-
ships," to cite the outline. As for our confidence in this way of formu-
lating stipulations, I can only say that it is akin to the confidence of a
parent, who has watched his child sit up, crawl, and then walk, that
someday he will talk. At the current stage of development, our con-
fidence must lie in the expectation of a "normal" course of growth,
within which we are at a recognizable stage.

During the course of growth, however, it is difficult to know how far
growth will proceed. There is a great discontinuity between thinking
about research as a publicly conducted need-meeting inquiry and think-
ing about the organized product of research. Are mathematically ex-
pressed function forms, which we tend to accept as the scientific last
word, really the ideal of social psychology? Or are they simply con-
veniences for effective testing of hypotheses? Is social science, like physi-
cal science, capable of meaningful formulation in the language of
mathematics? Or will it require the development of some other lan-
guage? Lewin, for example, felt that a nonmetric mathematics, hodology,
would be most appropriate for dealing with social-psychological phe-
nomena. We might conclude that the language of social science will
be mathematical but new kinds of mathematics may have to be devised.
Certainly we mathematical amateurs can sense that some recent devel-
opments are reassuring. The work of the Center for Mathematical
Biology at Chicago in elucidating mathematical theory of organisms,
the studies of the cybernetics people in taking account of feedback,
possibly, the research of the Rand Corporation in its pursuit of stochastic
applications to decision process— all these hold out hope that a useful,
rather than merely a formal science can be developed. Meanwhile, the
best apparent strategy is to continue studying human beings, to formu-
late the best backlog of propositions that we can, and try to get into
communication with the more approachable theoreticians among the
mathematicians. Certainly there seems no good reason for confining our
studies to the sorts of things that mathematical amateurs can handle
with correlation and covariance.

I have indicated that our intervening variables enable us to move
from one domain to another: from individual to group, for example.
Such variables are necessary to the extent that we view human enterprise

600 HERBERT A. THELEN

in terms of part-whole relations, in which every part of a whole is also
a whole made of smaller parts. The terms independent-intervening-
dependent apparently refer to the course of investigation rather than to
the structure of the system. We may find better terms for the latter. To-
ward this end, terms such as "input, output, internal, and external" may
be suggestive.

MENSURATIONAL AND QUANTIFICATIONAL PROCEDURES

Our quantification is actually prequantitative in every sense except
that there is usefulness in the simple ideas of "more than" and "less
than.53 We note, for example, three instances of two-level work during a
particular minute within a meeting. Literally, this means that at three
different times the observer thought he had enough evidence to justify
such a rating. It is the observer's opinion that we count, and aside from
the fact that he probably has different degrees of confidence in his three
judgments, it may be that wre could call each judgment a "unit.53 (As
a matter of fact, it works surprisingly well. ) Yet, speaking rigorously, we
are on shaky ground. Do all the three contributions have equal influence
on the group? Are they all equally valid indexes of the group state of
affairs? Probably not.

We could ask similar questions about our weightings of four "quali-
ties" of work. Again, the quantitative aspect is simply one of ranking —
fourth-level work seems to embody most fully the conscious seeking of
"reality" that the concept of "work55 connotes. Third-level work is a less
complete realization, second-level even less, and so on. But nobody
would seriously claim that a fourth-level statement is "worth" two
second-level statements. A further source of difficulty is the fact that we
count frequencies, which we interpret as intensities.

These problems arise from the interesting fact that our measuring
instrument is the nervous system of the observer. Someday, neurology
may have much to offer our ideas about quantification. Meanwhile, we
shall continue using our current procedures.

The biggest problem of quantification is also the biggest problem of
conceptualization: we think we are studying phenomena "out there,"
objectively, which we cannot do; or we talk as if we were studying our
own reactions to the phenomena, which is partly true. But what we need
is some way to talk about the interaction between observer and phe-
nomena— -for that is what we are really dealing with. If this is the case,
should we try to find that imaginary line between observer and phe-
nomena, and substract the observer from the total? This would give an
"objective" science, but what would it mean? Could somebody else come
along and use this "objective" science objectively, or would he already

Work-emotionality Theory of the Group as Organism 601

be interacting with the situation so that he must add a liberal dose of
self-knowledge to the science in order to be able to reach any reasonable
predictions or explanations? I can offer no reply to these questions. Yet
they point to what may be a matter of crucial importance; namely, in
social psychology we always study phenomena from a position of in-
volvement in them, and this is fundamentally different from our ex-
ternal position with respect to physical phenomena.

FORMAL ORGANIZATION OF THE SYSTEM

First among the ideas comprising the system are the postulates, ideas
that guide the processes of research. These postulates, introduced at the
beginning of our discussion of the Structure of the System as Thus Far
Developed, are the policies by means of which relevant factors are sifted
from irrelevant and attention is directed to the problems of research
method. The postulates show us how we must go at the job of investiga-
tion, and they suggest the methodological problems that we shall have to
solve. They help us avoid such fallacies as looking for the lost coin under
the street light simply because there it is light enough to see.

When used in this way, postulates specify the nature of the inter-
action among researcher, phenomena., and classes of concepts. We de-
cide, for example, to act as if interaction were sequential and contained
distinguishable "natural units." This notion has apparent validity in the
sense that it seems to fit our own past experience; it "feels" right and we
can already point to things we have experienced that seem to bear it
out. It is not, however, a hypothesis because it is not subject to test; it is
actually a criterion of method. We must seek until we can find a method
of research which identifies "natural units" within the sequence. If a
method docs not result in the finding of natural units, the method is
abandoned as unsatisfactory; it somehow is not appropriate to our in-
ternalized understanding of the nature of the phenomena we are dealing
with.

A glance at the postulates will show that we set ourselves a rather
knotty set of methodological problems to solve: how to discover "hidden
agendas" ; how to classify and record "feelings" ; how to diagnose stresses
to which individuals are susceptible; how to view the group as an "in-
teractive network" even though all the observable behaviors are produced
by individuals; how to conceptualize the "interaction between personality
and group." Postulates are dredged up from within the prior experiences
of the researcher, and they serve as axioms capable of endless elaboration
in a large number of experiments. They tell us which ways of proceeding
are fruitful and which will not pay off; and they give continuity to. di-
verse investigations over the years. They are formulated through con-

602 HERBERT A. THELEN

scious effort to become aware of the meaning of one's biases and pre-
dilections with respect to research designs.

The propositions, on the other hand, are not instructions to the re-
searcher so much as a portrayal of the broad fabric of human experience.
They spell out the "whole" of which the situation studied is a "part";
or they delineate the "ground" against which the universe studied is
"figured." Propositions are high-level abstractions which spell out the
nature of the phenomena being studied as seen from "within" the phe-
nomena rather than from the point of view of the observer. Perhaps the
distinction between postulate and proposition can be made clearer by
noting that the postulates imply the researcher's purposes, whereas the
propositions imply the rationale of the actors being studied. For us, a
single proposition makes no sense; the whole set is required to com-
municate the conceptual orientation within which theory is to be de-
veloped. Propositions communicate the nature of the relationships being
studied; they provide the "form" which is then fleshed out in theory.

Our "theoretical" statements are conceptual definitions of the vari-
ables, especially the "independent" and "intervening" variables discussed
earlier. These are related to behavior through their operational defini-
tions. The same variables are studied in all situations, but their opera-
tional definitions may vary from situation to situation in the sense that
different behaviors may be counted as evidence for the same variable.
Thus there is a large number of behaviors which we call "flight" and in
any particular situation of flight only a few of these behaviors appear.

In our work, the body of propositions is rather more coherent than
the collection of theoretical generalizations produced directly from re-
search. The reader may believe that actually we are rather more in-
terested in the propositions than in the theories, that we regard our
theories mostly as statements that help us spell out and modify our
propositions. I do not know how sound this would be, but I am con-
vinced that it is the propositions rather than the theories that men live
by. It is propositions rather than theories which become embedded in
cultures and thus determine how a given people at a particular time and
place will govern their lives. For me, the elegance, difficulty, and tech-
nical virtuosity of theory is an object of appreciation but the internalized
and homely understandings communicated in propositions represent the
more fundamental and significant contact with "reality."

SCOPE OR RANGE OF APPLICATION OF THE SYSTEM

We shall consider that the scope of the system is the same as the
scope of the propositions about groups presented earlier. The range can
be specified by mentioning the kinds of problems the system can deal

Work-emotionality Theory of the Group as Organism 603

with. The first problem, and the one on which most experimental work
has been done, concerns the relationships between individual personality
and behavior as a member of a group ; the system is especially adequate
to predict and explain these connections. The second problem derives
from the first: the performance characteristics of groups composed of
different combinations of personalities. The third problem is concerned
with the formation of subgroups and the explanation of group dynamics
as interaction among the subgroups. The fourth problem is the diag-
nosis of "hidden" agendas and buried purposes within the group. For
these types of problems, the methodology is complete; we know how to
proceed to explain and, within limits, to make a variety of specific pre-
dictions.

As explained earlier, the methods and experiments have been mostly
concerned with one kind of group, the "human relations training group."
Such groups were used because in the normal course of their business
together the members produce a great deal of explicit information about
their feelings, the problems of the group., and other relevant content.

There is no reason why the problems described above cannot also be
dealt with in respect to all kinds of meetings of all kinds of groups. The
differences of applicability of the system would come from problems of
accessibility of needed data, not from theoretical difficulties.

There are two areas of application which have been tested in ex-
perience, but not formally experimented with. The first is the creation
of methods and procedures for groups to use to achieve their own par-
ticular purposes. This area of application is suggested because the diag-
nostic methods are useful for understanding the inner and outer demands
that groups must handle. Thus, the research concepts have been prac-
tically useful in the creation of a citizen movement for rebuilding the
Hyde Park community in Chicago [50, 53]. More recently, the ideas
have played a major role in the creation of a "new" program for train-
ing elementary school teachers at the University of Chicago [49].

The second area of application is the study of cultural differences.
At the present time, for example, I am one of a team engaged in
traveling about Europe and organizing workshops in human relations.
Our training "philosophy" and methods are the same in various coun-
tries, and in most we have two weeks, full time, to work with one or
more groups. In effect, we are a uniform probe (or irritant) injected
into each country; and the diagnostic elements of the system have en-
abled us to define in a preliminary way a great many differences among
the groups of a sort that are usually thought of as reflections of differ-
ences in "national culture [48]." In other words, when studied inten-
sively, the actions of a group may throw considerable light on the basic
assumption of the larger culture.

604 HERBERT A. THELEN

It should be clear by now that there are many overlappings in ap-
proach among the various systems now being developed. Work on per-
ception, sociometrics, leadership, and group composition may all be
easily assimilated in the framework of propositions we have discussed.
We need to reach out in three directions: first, into more elaboration
of the external system which, at present, is represented mostly in the
categories of work and in some of the hidden problems of the group.
We have tended to pay close attention to "process" and to use such
structural concepts as were needed to describe the situation within which
the processes were occurring. A more sociological approach is to de-
scribe the structure as fully as possible, and call on the psychologist only
as needed to understand some of the dynamics associated with strains in
the structures. These approaches begin at different ends of the phenom-
ena, and they should be pulled together.

Much new development is also required to nail down the demands
of tasks. Very little work has been done in any systematic way for the
purpose of finding basic dimensions for classifying the task-demands
a group must deal with in order to achieve its purposes.

The third direction, in this case for more reaching out as well as
for further development within the system, is in the explication of the
intermember "identifications" which are at the psychological heart of
"groupness." We need to assimilate more psychoanalytic thinking within
our generally psychiatric approach.

HISTORY OF SYSTEM TO DATE IN MEDIATING RESEARCH

Several researchers — mostly working for the doctorate — have found
a place in the developmental program and have made substantial con-
tributions to it. Most of their original work is incorporated in disserta-
tions In the department of education and psychology, and in the Com-
mittee on Human Development at the University of Chicago, but some
studies are abstracted in the two monographs [44, 54] referred to in the
opening paragraph of this paper; the monographs also include a few
additional studies.

It is clear in retrospect that the various researches often initiated
in seeming independence from each other, fit together within a develop-
mental series of investigations. This fact derives both from the influence
of the student's major adviser and from the influence of the "culture"
of the Human Dynamics Laboratory: even when working separately,
the students have generally identified themselves as the staff of the
laboratory. Although their needs and interests have been different, they
have tended to speak a common language, to work together in training

Work-emotionality Theory of the Group as Organism 605

and workshop situations, and to serve as a manpower pool on which re-
searchers could draw for skilled help when needed.

In 1944, Thelen [46] compared the learning of freshman chemistry
classes. Half the classes did their laboratory work from typical laboratory
manuals; the other half planned their own experiments under the
guidance of the teacher. Although not a central object of inquiry in this
experiment, the greater self-direction and efficiency in the laboratory of
the "planning" groups was noted, but this was primarily a result of
better cognitive orientation and prerehearsal, during planning, of the
experimental manipulations.

In 1948 arid 1949, the experiments of Withall [56], Flanders [15],
and Perkins [36] demonstrated the importance of affective communica-
tion between teacher and student. Affect was seen as a concomitant
of the ' 'intent53 of the communicator, and was found to influence recall,
anxiety, perceived "feeling,35 and a number of physiological measures.
During this period, Rehage [38], in an experiment involving teacher-
class planning, noted the importance of the teacher3s response (or lack
thereof) to feelings expressed by students; and he also found marked
development of cohesion in the sociometric pattern of a class which had
shared intense feelings together.

The effort to understand why teachers (and other leaders) re-
sponded as they did to the feelings of others led Glidewell [17] to the ex-
perimental study of interpersonal anxiety as related to the behavior of
the leader. The four members of the group were trained to play the roles
of people who had been identified as either anxiety-producing or anxiety-
allaying through clinical study of the leader. They were trained to make
"supportive3' or "threatening33 statements, and observers correctly spot-
ted these two types of statements by noting their consequences in the de-
terioration or "strengthening33 of the subject's style of leadership.

About this time (1950) Thelen [47] published a methodological
analysis of the postulates required for research on groups, and this fore-
cast with some accuracy the nature of succeeding work in the laboratory.

In the same year, deHaan [13] attempted to use Bion's emotion and
work concepts as a basis for sequential plotting of group interaction. The
results, though crude, were encouraging and led to considerable further
effort at refinement and systematization of the method. During this pe-
riod there were several other tests of the usefulness of the concepts of
emotion and work tendencies in personality. Thus Stock [43] had fair
success in predicting sociometric choices; D. McPherson [34] showed
that an individual emotion-work sentence completion test was more
useful than the TAT for predicting emotion and work behaviors in the
group. In connection with some research with the Air Force (1952),
J. McPherson [35] showed relationships between e-w personality pat-

606 HERBERT A. THELEN

terns and tendencies to distort the meaning of close-to-self written mate-
rials; B. Sarchet [39] used the same personality data to predict roles of
members in groups of officers. Working under the same contract. Glide-
well [18] showed that certain characteristics of the solutions to prob-
lems, worked out by 50 different 12-man groups of officers, could be
differentially predicted from knowledge of the e-w patterns of the groups
and from knowledge of the groups5 standards controlling the expression
of feeling (e.g., labile, constrictive, integrative). These notions of con-
trol, incidentally, have yet to be worked into the over-all system. In the
same year, Freedman [16] studied the way eight different teachers dealt
with emotionally charged discussion following presentation of a provoca-
tive standard dramatic story of their classes; and he related their "styles"
both to teacher anxiety and to lack of congruences between the percep-
tions of the teacher and the students.

Beginning in 1951, the experimental program summarized in the two
monographs was begun under the auspices of the Group Psychology
Branch of the Office of Naval Research. This work was brought to a
close officially in July, 1956. The first monograph [54] reports the devel-
opment and validation of the basic e-w assessment instrument (the Re-
action to Group Situation Test) by the Gradolphs, Stock, and Hill,
the development of the method of sequential analysis and interpretation
by Stock and Thelen, the development of an objective method for
"unitizing" interaction by Ben-Zeev, and the method for studying inter-
subgroup dynamics by Stock and Hill. The second monograph [44]
gives the substantive findings of Stock, Gradolph, Hill, Glidewell, Lieber-
man, and Mathis with respect to behaviors of individuals in groups,
group composition, "trainability," and productivity. In 1956, Thelen
[52] published a preliminary over-all statement of the propositions and
aims of the research.

As this is written, Stock and Lieberman are applying the system to
the study of focal conflicts in therapy groups and Hill is studying growth
and development of therapy groups. These workers are located in VA
hospitals and training centers.

Probably, these are the aspects of the system which have had most
to do with its usefulness for instigating research :

1. It is still developing, so that there is a continual challenge to
creativity.

2. Its scope is such that students can use it in a wide variety of situa-
tions and for a wide variety of purposes.

3. It deals with problems and uses ideas which are personally in-
volving and exciting.

4. The research has tended over the years to be seen as a team
operation, and students have had a feeling of "place" in the program.

Work-emotionality Theory of the Group as Organism 607
EVIDENCE FOR THE SYSTEM

The evidence for the system as a coherent body of useful ideas is
probably of three sorts. Firstly, we can make valid predictions with it.
The fact that some of the operations of prediction cannot be carried
through as yet in the public domain does not operate against the sys-
tem; it does mean that the theory needs further development. Secondly,
as trainers in groups we habitually employ the concepts to help guide
the training process, and we find the concepts useful. Thirdly, the major
ideas of the system can be communicated to students and they can use
them for more effective participation in groups.

I am not aware of any incompatible data. So far we have assumed
that an encounter with such data meant merely that our concepts needed
further refinement, not that the data were embarrassing. For example,
when our initial predictions, using the concept of valency, were not borne
out, we discovered that we were using the term in three different senses.
Once these had been made explicit and used appropriately, predictions
were vastly improved.

The most critical test of the system will be through the method of
blind prediction described as experimental design 5, under Structure of
the System as Thus Far Developed.

No other system I know gives as adequate or useful a picture of
group process: hidden problems, transitions and phases, group climates,
and emotional phenomena in the group. Some other systems, such as
Homans's, seem more complete in offering a better over-all view of
group life in relation to institutional and community factors. Our sys-
tem is stronger in relation to personality factors.

EXTENSIBILITY OF METHODS AND CONCEPTS,
PROGRAMMATICITY, AND STRATEGY FOR

DEVELOPMENT OF THE SYSTEM

The work on the system has, if anything, strengthened our initial
"convictions" and postulates as discussed earlier under Initial Evidential
Grounds for Assumptions and Formal Organization of the System. Hence,
I feel that these ideas may be fundamental to social psychology. I further
think that the use of sequential analysis as an aid to the diagnostic proc-
ess is worth considering seriously in a wider range of behavioral studies ;
and that the frank recognition of the use the researcher makes of his
own nervous system might help unleash more creativity.

The system is moving, I think, toward a broad generalized view of
human behavior. It will be especially interesting to see how far the
group concepts here experimented with can usefully apply to internalized

608 HERBERT A. THELEN

groups within the individual. If they can, then the system will expand
to include some new ideas about individual psychology.

The program has been realized almost completely with respect to the
research methodology, although simplifications may be possible; it is
realized in broad outline as far as the underlying propositions are con-
cerned, although extension is needed into more sociological and psycho-
analytical concepts. So far as theory development is concerned it is only
partially realized. A great deal of effort will be needed within a
systematic program to work out and explicate the numerous cross rela-
tions possible among the variables.

With regard to the development of psychology in general, I hesitate
to comment on the chief barriers to theoretical advancement. I have
sought to make clear the kind of thinking and method that seem essential
to me. Other people, with other needs, can make progress with other
methods, and should be encouraged to do so. Nevertheless, the following
general problems must be dealt with, possibly with fresh approaches, if
we are to advance to new levels of reorganization and integration of
ideas :

1. The problem of free will vs. determinism. Science must assume
a deterministic position, but men resent, for good reason, the notion that
they are without choice. We must learn how to work into our systems
a variety of concepts taking due account of feedback, of learning, and
of creative emergence into consciousness of "new" ideas. I find con-
siderable encouragement in the failure of certain recent applications of
stochastics to the decision process. These failures underscore the sig-
nificance of this problem.

2. The problem of "probability35 states. Our designs should stick
closer to the facts of behavior: instead of comparing probability of our
findings (significance) against "chance" or against regression from
initial conditions, we should formulate several possible final states and
assess their relative probabilities under the circumstances of the experi-
ment.

3. The problem of developing mathematical functions more ap-
propriate to studies of human interaction.

4. The problem of dealing with interactions between the researcher
and the phenomena he studies.

REFERENCES

1. Bales, R. F. Interaction process analysis. Reading, Mass.: Addison-
Wesley, 1950.

2. Ben-Zeev, S. The construction and validation of a method for unitiz-
ing sequences of group interaction. Unpublished doctoral dissertation,
Univer, of Chicago^ 1955. Abstracted in [56].

Work-emotionality Theory of the Group as Organism 609

3. Bion, W. R. Experiences in groups. I. Human Relat., 1948, 1, 314-
320.

4. Bion, W. R. Experiences in groups. II. Human Relat., 1948, 1, 487-
496.

5. Bion, W. R. Experiences in groups. III. Human Relat., 1949, 23 13-
22.

6. Bion, W. R. Experiences in groups. IV. Human Relat., 1949, 2, 295-
303.

7. Bion, W. R. Experiences in groups. V. Human Relat., 1950, 3, 3-14.

8. Bion, W. R. Experiences in groups. VI. Human Relat., 1950, 3, 395-
402.

9. Bion, W. R. Experiences in groups. VII. Human Relat., 1951, 4, 221-
227.

10. Blocksma, D. Leader flexibility in group guidance situations. Educ.
psychol Measmt, 1949, 9, 3.

11. Gartwright, D., & Zander, A. (Eds.) Group dynamics. Evanston,
111.: Row, Peterson, 1953.

12. Gattell, R. B. Concepts and methods in the measurement of group
syntality. Included in [19].

13. deHaan, R. Graphic analysis of group process. Unpublished doctoral
dissertation, Univer. of Chicago, 1 95 1 .

14. Deutch, M. A theory of cooperation and competition. Human Relat.,
1949, 2, 129-152.

15. Flanders, N. A. Personal-social anxiety as a factor in experimental
learning situations. /. educ. Res., 1951, 45, 100-110.

16. Freedman, M. The effect of the teacher's role on the group's will-
ingness to participate in problem-solving. Unpublished doctoral dissertation,
Univer. of Chicago, 1952.

17. Glidewell, J. C. Prediction of some aspects of group leadership be-
havior. Unpublished Master's essay, Univer. of Chicago, 1949. See also:
The teacher's feelings as an educational resource. /. educ. Res., 1951, 45,
119-126.

18. Glidewell, J. C. Group emotionality and productivity. Unpublished
doctoral dissertation, Univer. of Chicago, 1953.

19. Hare, A. P., Borgatta, E. F., & Bales, R. F. (Eds.) Small groups.
New York: Knopf, 1955.

20. Hill, W. F. The influence of subgroups on participation in human
relations training groups. Unpublished doctoral dissertation, Univer. of
Chicago, 1955.

21. Homans, G. C. The human group. New York: Harcourt, Brace,
1950.

22. Horwitz, M., & Cartwright, D. A projcctive method for the diag-
nosis of group properties. Human Relat., 1953, 6, 397-410.

23. Jennings, Helen H. Leadership and isolation. (2cl cd.) New York:
Longmans, Green, 1 950.

24. Jennings, Helen H. Sociometric differentiation of the psyche group
and the socio group. So dome try, 1947, 10, 71-79.

610 HERBERT A. THELEN

25. Korzybski, A. Science and sanity. Glencoe, 111.: Free Press, 1936.

26. La Barre, W. The human animal. Chicago: Univer. of Chicago
Press, 1954.

27. Lewin, K. Frontiers in group dynamics: concept, method, and
reality in social science: social equilibria and social change. Human Relat.y
1947, 1, 5-41.

28. Lewin, K. Field theory in social science. New York: Harper, 1951.

29. Lewin, K. Resolving social conflicts. New York: Harper, 1948.

30. Lewin, K., & Lippitt, R. An experimental approach to the study
of autocracy and democracy: a preliminary note. Sociometry, 1938, 1,
292-300.

31. Lieberman, M. Two studies included in [44], in press.

32. Marks, G. Time-lapse photography in the study of group process.
Unpublished Master's essay, Univer. of Chicago, 1949. See also [47,
chap. 4].

33. Mathis, A. Development and validation of a trainability index for
laboratory training groups. Unpublished doctoral dissertation, Univer. of
Chicago, 1955.

34. McPherson, Dorothy. An investigation into the nature of role con-
sistency. Unpublished doctoral dissertation, Univer. of Chicago, 1951.

35. McPherson, J. A method for describing the emotional life of a group
and the emotional needs of group members. Unpublished doctoral disserta-
tion, Univer. of Chicago, 1951.

36. Perkins, H. V. Climate influences group learning. /. educ. Res.,
1951, 45, 115-119.

37. Redl, F. Group emotion and leadership. Psychiatry, 1942, 5? 573-
596. Included also in [19].

38. Rehage, K. J. A comparison of pupil- teacher planning and teacher-
directed procedures in eighth grade social studies classes. /. educ. Res.., 1951,
45, 111-115.

39. Sarchet, Bettie. Prediction of individual work roles in two adult
learning groups. Unpublished doctoral dissertation, Univer. of Chicago,
1952.

40. Scheidlinger, S. Psychoanalysis and group behavior. New York:
Norton, 1952.

41. Steinzor, B. The development and evaluation of a measure of social
interaction. Human Relat., 1949, 2, 103-121.

42. Stephenson, W. The study of behavior. Chicago: Univer. of Chicago
Press, 1954.

43. Stock, Dorothy. The relationship between the sociometric structure
of the group and certain personality characteristics of the individual. Un-
published doctoral dissertation, Univer. of Chicago, 1952.

44. Stock, Dorothy, & Thelen, H. Emotional dynamics and group
culture. Washington, D.C.: National Training Laboratories, 1957, in press.

45. Stogdill, R. M. Leadership, membership, and organization [11,
chap. 4].

Work-emotionality Theory of the Group as Organism 61 1

46. Thelen, H. A. A methodological study of learning chemical con
cepts. /. exp. Educ., 1944, 13, 26-78.

47. Thelen, H. A. Educational dynamics: theory and research. /. soc.
Issues, 1950,6 (2), 1-95.

48. Thelen, H. A. European groups, a challenge to American education.
Elementary Sch. J.9 April, 1957, 351-362.

49. Thelen, H. A. Teacher preparation in the future. In 1957 Yearbook
American Association of College Teachers of Education, in press.

50. Thelen, H. A. Dynamics of groups at work. Chicago: Univer. of
Chicago Press, 1954.

51. Thelen, H. A. Five lectures on group dynamics presented to British
Institute of Management, 1956. (Mimeo.)

52. Thelen, H. A. Emotionality and work in groups. In L. White (Ed.),
The state of the social sciences. Chicago: Univer. of Chicago Press, 1956.
Pp. 184-200.

53. Thelen, H. A., & Sarchet, B. B. Neighbors in action. Univer. of
Chicago, Human Dynamics Laboratory, 1955. (Planographed.)

54. Thelen, H. A., Stock, Dorothy, Ben-Zeev, S., Gradolph, L, Gradolph,
P., & Hill, W. F. Methods for studying work and emotionality in groups.
Univer. of Chicago, Human Dynamics Laboratory, 1954. (Planographed.)

55. Thelen, H. A., Stock, Dorothy, Ben-Zeev, S., Gradolph, I., Gradolph,
P., £ Hill, W. F. The application of Bion's theories to the study of small
groups, 1953. Pp. 1-62. (Dittoed.)

56. Withall, J. The development of a technique for the measurement
of social-emotional climate in classrooms. /. educ. Res., 1951, 45, 93-100.

AN APPROACH TO PSYCHOLOGICAL

THEORY IN TERMS OF THE

THEORY OF ACTION TALGOTT PARSONS'

Harvard University

Parti
Introduction

Background Factors and Orienting Attitudes {1} 619

Background factors 619

Orienting attitudes 624

Part II

The General Theory of Action and Its Application to Psychological Systems

{2,3,6, 7} 627

Structure of the theoretical system 627

Psychological systems 644

Psychological System and Organism 647

The Object-relations of Psychological Systems 651

Psychological and Cultural Systems 656

The Internal Structure and Processes of Psychological Systems . . . . 659

The Internal Differentiation of Psychological Systems 672

A Summary of Psychological Problems 679

Some Levels of Organization of Psychological Systems 681

Structural Change in Psychological Systems 688

1 1 should like to acknowledge a special obligation to Dr. James Olds. I had
hoped to enlist him as coauthor, but geographical separation kept us from making
it a genuine joint product. Nevertheless, besides previous collaborative work, I
did enjoy the benefit of a few days' discussion with Dr. Olds in June, 1955 (and
a working paper he wrote subsequent to this discussion), and two further days in
April, 1956. Partly because of the thinness of my own command of the literature
of experimental psychology, but even more because of its intrinsic excellence, I
have leaned heavily on Olds's recent book, The Growth and Structure of Motives
[20]. I should, therefore, like to consider Dr. Olds as "special consultant.5' He
should not, however, in any way be held responsible for the views expressed here.
I am also particularly grateful to Dr. Anne Parsons who carefully edited the
manscript in the interest of style and clarity and contributed to a number of
substantive problems.

612

Psychological Theory in Terms of Theory of Action 613

Part III

Some Questions of Methodology and of the Scientific Significance of the System 690

Initial evidential grounds for assumptions of the system {3}- .... 690

Construction of function forms {4} 692

Mensurational and quantificational problems {5} 696

Formal organization of the system {6J- 700

Scope or range of application of the system {7} 702

Evidential status; prospective considerations {8-12} 707

References 709

Part I

INTRODUCTION

This essay is concerned with a rather special type of psychological
theory; one which is expressly treated as part of a more general con-
ceptual scheme embracing the processes of social interaction and the
patterns of culture as well as the traditional subject matter of psy-
chology. The term "general theory of action" has been used to designate
the wider scheme which, in turn, can be broken down into several dif-
ferent parts or subtheories.

The most fundamental of these breakdowns is based on four reference
points: organism, personality or psychological system, social system, and
cultural system. All four are abstractions from and modes of analyzing
the phenomena of the behavior of living organisms. They are not con-
cretely, only analytically, separable. The system's central, but by no
means exclusive, interest is in human behavior.2

The same concrete behavior usually involves all four reference points.
In a sense to be explained later, the four interpenetrate each other.
Yet the analytically articulated systems which we have isolated for
theoretical analysis and for defining the relevant empirical data are not
mutually reducible; the discrimination of the four systems is not merely
tautologous.

Action constitutes systems, which involve the relations of one or
more actors (i.e., behaving organisms or parts of them or collectivities

2 From some points of view the terms action and behavior may be treated as
interchangeable. One may suggest, however, that behavior be the term applied to
the total complex of obscrvablcs in a given case, action the term which includes
both the observables and the theoretically postulated intervening variables and
"covert" processes. It will also be noted that, compared to previous publications,
we speak of four rather than three primary subsystems of the general theory of
action; the organism., in certain aspects, has been added to personality, social
system, and culture. This represents a definite thorcticai innovation.

614 TALCOTT PARSONS

involving a plurality of them) to a situation, composed of other actors
and/or "nonsocial" objects. The systems concerned are always con-
stituted by the relations between one or more actors and one or more
objects in its or their situation. This is a fundamental point: the
actor is not conceived as one system, which acts in relation to a situation
(or environment) which is then treated as another system; actor and
situation together constitute the system of reference. This is as much
the case for a psychological system, as a system, as for the other
types. A "personality" conceived as devoid or independent of "ob-
ject-relations" could not be called a system of action in these terms.
This is much the same as saying that behavior is the empirical subject
matter of the theory of action. The properties of a behaving organism,
independent of its behavior in actual situations, are of interest to that
theory only in so far as they condition or are otherwise involved in the
behavior.

According to this conception, a social system is a system generated
and constituted by the interaction of two or more individual actors,
whereas a psychological system is a system of action characterized by the
fact that all the behavior belonging to it is behavior of the same living
organism.3 Again, whatever properties the actors may have which are
independent of the processes of their specific interaction with each other
are no subjects for the analysis of social systems except in so far as those
properties bear on the interaction; i.e., are factors contributory to or in-
volved in it as resultants.

Thus by definition, all concrete behavior belongs to some psy-
chological system, and a very large part of it at the same time belongs to
some social system. Yet the same organism participates in a plurality of
social systems; conversely, the same social system — over a period of
time — may be "composed" of different behaving actors and yet remain
"the same system." The two are thus overlapping but also crosscutting
modes of organizing the data of behavior for scientific analysis.

A cultural system is a system which defines and maintains patterns
of the meanings of actions and of objects which function in the orienta-
tion of actors in psychological and social systems. Orientation is always
the patterning of the relations of one or more actors to one or more
objects in a situation. As a generalized mode of orientation, a cultural
pattern is at least potentially applicable to more than one object and
characteristic of more than one actor. Cultural patterns are transmissible

8 It is convenient to reserve the term personality strictly speaking for the total
behavior system of a given living organism; it is thus parallel to society rather
than to social system. Hence we propose that the term psychological system be
used as parallel to social system, and personality as parallel to society. In these
terms, subhuman organisms would certainly be defined as having personalities.

Psychological Theory in Terms of Theory of Action 615

from one empirical action system to another. As components of psy-
chological systems, they must be learned (from other actors or created
through learning processes in the system in question). They may be "dif-
fused" from one social system to another, or finally, embodied in physical
objects which function as signs or symbols. When conformity with the
"definition of the situation35 embodied in cultural patterns becomes an
aspect of the "structure of the system," we speak of them as coming to
be internalized in psychological systems, i.e., personalities, and in-
stitutionalized in social systems. The same cultural patterns are both
internalized and institutionalized.

In its relevance to the theory of action, the organism is that aspect
of the physiologically functioning system which interacts directly with
the personality and the other systems of action. It is the source of
energy for all processes of action and the source also of a complex of
essential facilities and rewards. Although based on a genetic constitution,
its own organization is substantially influenced by the processes of
conditioning and learning which occur in the life history of the in-
dividual. For many purposes, only part of the total concrete organism
should be treated as part of the system of action. Later we will refer
to this part as the "behavioral organism" and distinguish it as subsystem
other than the "vegetative organism."4

The general theory of action maintains that these four orders of
system, and various others which can be derived by analysis of them
and of their interrelations,5 should not be treated as independent of each
other except in the sense that differentiated parts of the same complex
of phenomena are partially independent. Theoretically, the analytical
schemes appropriate to the different systems should be derivable from a
common set of postulates and definitions of fundamental variables and
relations. Each subtheory should depend on parametric considerations
which define empirical constants in which the same fundamental
variables operate. It is thus necessary to differentiate within the general
framework different classes of system and to relate these different systems
to each other.

Within each of the four basic types of system, the applicability of
the theoretical scheme is not limited to one particular "level" in the
microscopic-macroscopic range. In social systems it applies all the way
from the small experimental group to the large-scale society; in psy-
chological systems, from a single stimulus-response pattern to the total
personality; and in cultural systems, from the specialized set of "under-
standings" of a married couple, for example, to the total culture of a

4 As this term has been used by Franz Alexander. Cf. [1].

5 For example, we treat economic theory as dealing with a special type of
social system, an economy. Cf. [27].

616 TALGOTT PARSONS

major civilization. In this crucial respect, the relation of the theory of
action to the relevant empirical systems resembles that of mechanics
which explains falling apples and the motions of the planets by the same
set of laws.

We have noted that these four primary subsystems of action are not
mutually reducible. No one of them has ontological priority in the sense
that the laws worked out in relation to it have only to be "applied33 to the
less fundamental levels. On the other hand, the four systems are not
arbitrarily juxtaposed so that the order of their relations does not matter.
On the contrary there is a quite definite order, clearly an order of levels
of organization and control. As seen in this order, psychological systems
organize and control the behavioral organism, social systems organize
and control psychological systems, and cultural systems organize and
control social systems. Looked at from the opposite perspective the order
is one of "conditions." Social systems provide the most immediate set
of conditions on which the functioning and development of cultural
systems depend, psychological systems provide a set of conditions under-
lying the functioning of social systems, and the organism provides condi-
tions underlying psychological systems.

We postulate a complete continuity between biological systems and
systems of action; from this point of view, action is a specialized aspect
of life. It is essentially that aspect in which life processes transcend the
internal mechanisms of the individual organism and the metabolic
interchanges with the environment. The starting point for action is the
organic differentiation of perception and locomotion from other func-
tions and the consequent enlargement of the range of adaptation made
possible, especially through control by the central nervous system.

The relative importance of the organic, psychological, social, and
cultural factors is a function of stages or levels of the evolutionary scale,
but in the order just cited the later terms grow more prominent as we
move up the scale. All of them are discernible below the human level.
Some kind of learning is apparently found well down the evolutionary
scale and rapidly becomes more significant with organic development.
However important the genetic constitution of the organism, however
important, during maturation, may be the "unfolding33 of behavioral
capacities through the operation of genetically determined mechanisms,
behavior comes increasingly under the control of systematically organized
learned processes. With respect to these, organisms of the same genetic
constitution may differ, but within the species there are uniformities
determined by relatively uniform conditions of learning. Learned be-
havior is the focus of what we mean by psychological systems.

With respect to learning, psychological systems originate in the
relations of the organism to the total environment. A focal point, how-

Psychological Theory in Terms of Theory of Action 617

ever, is interaction with other organisms in which the behavior of the
other organism becomes an essential aspect of the determinants of
learned behavior for the organism of reference. Essential functions for the
individual organism and for the species thus come to be dependent on
the effective regulation of these interactive processes. A crucial intra-
species case is sexual reproduction which is never exclusively regulated
by the triggering of instinctive patterns of behavior even at the sub-
human level but always involves some psychosocial regulation of the
relations of the partners. To deny this would imply that on the en-
vironmental side meetings, including all the detailed conditions of suc-
cessful joining, were exclusively a matter of chance encounters.6

If primary biological functions are dependent on social interaction,
there must be mechanisms by which the behavior of interacting organisms
is somehow made to match, so that there is a probability greater than
chance that each, in response to the other, will come to perform the
appropriate kinds of acts. When a plurality of interacting organisms tend
to interact in systematically organized ways in relation to each other
as the result of learning, we may speak of a social system.

What is spoken of as perception is clearly a matter of some kind of
organized sensitivity to environmental conditions. It is not the same as
the more simply "reactive" sensitivity of, say, the skin to prolonged ex-
posure to strong sunlight, or the respiratory apparatus to pronounced
diminution of the oxygen content of the atmosphere. The distinctive
feature of perception is the reaction of the organism to stimuli, organized
with reference to environmental events. But when social interaction
appears, a further level of the generalization of the meaning of such
events is added. The behavior of other organisms — and qualities which
become associated with their behavior — come to be interpreted as "in-
tentional" signs which guide the behavior of the organism of reference.7

The generalized patterning of the meanings of environmental objects
and events is the focus of what we here mean by culture. In its most
elementary forms it is not dependent on social interaction — Tolman's
cognitive map [cf. "A Psychological Map," 22], for example, is a
"cultural" factor in behavior. But only systems of social interaction
provide the conditions necessary for the most distinctive phenomena we
associate with culture, notably the transmission of systematically pat-
terned meanings from organism to organism without each undergoing
independently the original learning experiences by which the meaning-
pattern was established. We can clearly speak of sign-behavior, meaning,
generalization and communication on subhuman levels. But the organiza-

0 Important evidence on these processes will be found in [33].
7 A classical study of a rather elementary form of socially interactive sign
behavior is that of von Frisch on the behavior of bees.

618 TALCOTT PARSONS

tion and transmission of meanings, independent of experiential context,
seems on a large scale to be specifically human. The focal mechanism of
course is language.

Whatever the subhuman antecedents and prototypes, when the sub-
ject matter is human sociocultural behavior, there can be no doubt of
the involvement of all four of the system levels about which the present
discussion will revolve. When we deny that any one of the systems
higher in the order of organization and control is "reducible" to deter-
mination by a lower-order system, we mean that independent significance
must be attributed to the phenomena of organization on each level as
defined by: (a) selectivity of inclusion of and emphasis on components
available from lower-order systems, and ( b ) distinctive patterning of the
relations of the components selected.

Thus we assume that the same laws govern metabolic processes in
"vegetative" tissues, on the one hand, and in eyes, nervous tissue, or
great skeletal muscles, on the other. But tissues and organs which are
differentiated and specialized with reference to behavioral function in
the organism are not indistinguishable from those specialized with refer-
ence to vegetative function. The same biochemical-physiological com-
ponents are differently selected and organized. Similarly, in one sense,
the sensitivities to environmental influence which are brought together
under the heading of "capacity to learn" are not different from the
common features of the "irritability" of protoplasm. Within the con-
crete organism, however, learning becomes a specialized function in the
operation of which the proportions of the basic components are different
from those general to all organic functions. Further, effects of learning
as determinants of the subsequent behavior of the organism come to be
perpetuated. These two facts are not simply matters of the "average"
functioning of protoplasm, but involve a different selection and mode of
organization of the physiological components.

Again, other types of learning and that associated with social inter-
action, i.e., with continuing sensitivity to the behavior of other organisms,
have essential features in common. But the behavior focused on social
interaction comes to be differentiated from that centering on physical
objects alone. Among the general mechanisms of learned behavior there
is selectivity and special organization of those appropriate to the regula-
tion of social interaction. Finally there can -be distinctive selection and
patterning of relations where the primary focus is on a system of mean-
ings as such rather than on the processes of social interaction. The com-
ponents or "building blocks" out of which cultural systems are con-
structed are the same as those built into psychological and social systems,
but this does not mean that the systems themselves are identical

Psychological Theory in Terms of Theory of Action 619

The preceding discussion should have made it clear that the degree
to which it is important to discriminate among these different levels of
system analysis is by no means uniform for all problems. The relative
importance of discriminating seems to increase with the level on the
evolutionary scale which is being dealt with, and with the degree of
theoretical refinement to be reached. For many problems of animal be-
havior, it has not seemed very important to discriminate behavioral
organism and psychological system. Much of the content of the human
personality system is derived from social interaction, however; i.e., It
consists of "internalized" social objects. Hence, when we deal with
human personality, serious distortion may arise from the attempt to as-
similate this content to a paradigm of structure and function of the
organism independent of the structure of social systems.

Traditional social science has tended to rest content with discrimi-
nating between heredity and environment, in the sense of general bio-
logical theory and then, within the category of environment, distinguish-
ing the factors distinctive to the human environment. This has been the
key emphasis in the concept of "culture." In this sense, three of our four
systems are primarily "cultural"; for more refined purposes, it becomes
necessary to make further discriminations. I shall therefore use the label
"culture (I)" for culture as this very general designation of all factors
in human behavior except those determined by biological heredity and
the physical environment. Culture (II) then labels the sense in which I
have used it, which differentiates it both from psychological systems and
from systems of social interaction.

The foregoing discussion indicates the broadest frame of reference
within which I shall attempt to outline the main structure of the theory
of action and its application to psychological theory. Now, however, let
us turn briefly to the first rubric of the outline,

BACKGROUND FACTORS AND ORIENTING ATTITUDES

Background factors. Academically, the author of this essay began as
and still is a sociologist, not a psychologist. Like any person professionally
concerned with the scientific study of human behavior, from the be-
ginning I had a certain level of awareness of and concern with psycho-
logical problems. But this did not become what could be called a gen-
uinely technical interest until relatively late in my career.

That career began on the border line between sociology and
economics, but with considerable undergraduate background (at Am-
herst College) in biology and philosophy. Graduate study was at the
London School of Economics and at the University of Heidelberg,

620 TALGOTT PARSONS

Germany. My full commitment to sociology rather than economics did
not come until after graduate work — unlike an American doctorate, the
German program was sufficiently general to leave both doors open.

Within economics, my primary initial focus was what the 1920s
called "institutionalism," which naturally led to sociological interests.
These were reinforced by the atmosphere of the London School, and
even more, in Heidelberg, by acquaintance with the work of Max
Weber. My dissertation there dealt with a subject in the field of economic
institutions, the treatment of "capitalism" as an institutional system in
German economic literature, particularly by Marx, Sombart, and Max
Weber.

Interest in economic institutions evolved into interest in how lead-
ing economists had handled the sociological border lines of their the-
oretical problems, and conversely, how sociologists had handled the
problem of the place of economics in their thinking. On the economic
side this led to intensive study of the work of Alfred Marshall; on the
sociological, in addition to Weber, to Emile Durkheim, and Vilfredo
Pareto, a case spanning both disciplines. The outcome of this series of
studies was The Structure of Social Action [21], in which the conception
of a theoretical system first clearly emerged in my thinking. For all the
diversity of background and empirical interest, the work of these writers,
the book held, embodied a common conceptual scheme for the analysis
of social systems containing at least the beginnings of a generalized
theoretical system. Economics had such a generalized scheme; it seemed
to be the moment to search for an equivalent in sociology. The basis
for such a scheme could be quite different from the older "speculative"
evolutionary theories of which Herbert Spencer's was the prototype.

My focus in sociology was on the comparative treatment of in-
stitutions. As a result of the influence of Malinowski and Hobhouse
and Ginsberg in London, it also included a strong interest in social
anthropology and its treatment of primitive societies.

As yet, however, I had no genuinely technical interest in psychology.8
That first developed in connection with a set of problems concerning
the treatment of motivation in the traditions of economic theory. Its
key conception was that of the "rational pursuit of self-interest," the
most general available statement of what underlies various versions of
the "profit motive." I early became convinced that this could not be
treated in the usual sense as mainly a psychological generalization —
what has sometimes been called by economists a "propensity of human

8 On a more methodological level I was, however, considerably influenced
about this time by two psychologists, namely, Tolman, through his Purposive Be-
havior [35] and Kohler through the Mentality of Apes [14] and Gestalt Psychology
[15].

Psychological Theory in Terms of Theory of Action 621

nature" — largely because so-called capitalism is not an institutional
feature of all human society. Hence I decided to study medical practice
as an example of the contemporary professions where, on an ideological
level at least, it is categorically denied that economic self-interest does or
should be allowed to govern behavior.

One outcome of the study was definitely to confirm the hypothesis
that, although the doctrine of self-interest may be a valid empirical
generalization about motivation in modern business, it does not neces-
sarily hold for the professions. The difference, however, is not primarily
between types of motivation in a psychological sense, but rather between
institutional structuring of the situations in which the respective groups
act.9

In addition, a psychological interest which was both broader and
more technical emerged from the same study. Its starting point was the
"psychic factor in disease," as manifested in either psychosomatic or be-
havioral symptoms, a conception which was coming to be intensively dis-
cussed in medical circles (about 1935-1936). In this connection, I first
undertook careful and intensive reading of Freud and the work of other
writers in the psychiatric field.

This study confirmed my central view about occupational motivation,
above all because this type of psychology provided a clue essential to
understanding the functional basis of the institutional patterning of
medical practice. In analyzing the "irrational35 motivational factors in
the relationship of doctor and patient I became aware of their reciprocal
interaction on unconscious levels and the bearing of this on the pattern-
ing of their respective roles. These insights in turn widened into a general
interest in the problems of the relation between motivational structures,
broadly on the level on which Freud treated them, and the institutional
structure of the situation in which action takes place.

Various other writers, above all rny own colleagues Allport and
Murray, and W. I. Thomas and later G. H. Mead played important
parts in this development of psychological thinking and interest, but in
a variety of ways I kept coming back to Freud. This motivated seeking
as much psychoanalytic training as a nonmedical person was permitted
to acquire.10

Interest in developing relations between the "clinical" level of psy-
chological theory and the sociological analysis of institutional structure
was pursued for a considerable period and in a number of directions.
First there was an extension of the interest in medical practice as a social

0 The fullest report of the results of this study will be found in "Modern
Medical Practice," Chap. 10 of my later book, The Social System [22].

10 Under the Class C program of the Boston Psychoanalytic Institute. I
eventually became, and am now, an affiliate member of the Boston society.

622 TALGOTT PARSONS

phenomenon and its place in our society as a mechanism of social
control. Secondly there was an interest in kinship and family structure,
and their relation to the processes of socialization and social control.
Finally, there was an interest in the "social psychology" of certain mass
phenomena in their relations to macroscopic levels of the analysis of
social structure.11

This was a genuinely technical psychological interest, but a specialized
one in two respects. It was overwhelmingly defined by the ways in
which certain psychological materials fitted into the sociological problem-
contexts I have briefly outlined, i.e., their relations to the social system.
Secondly, the bodies of psychological theory which I studied most
thoroughly were those of Freud and certain derivatives from him (e.g.,
Murray, Kardiner, Homey, Fromm, etc.) and the special sociologically
oriented social psychology of Thomas and Mead. During this period, I
was less intensively concerned with matters of general theory. Never-
theless, the extension of theoretical synthesis beyond sociology to include
at least certain parts of psychology and of the "cultural" interests of
anthropology had been gradually taking place.

A critical set of steps occurred in connection with a program of
theoretical stocktaking which was carried out by various members of the
Harvard Department of Social Relations with the help of E. C. Tolman
and E. A. Shils as visiting collaborators in 1949-1950. One outcome of
this project was Toward a General Theory of Action [26], whose con-
tributor included, besides the editors, Shils and me, E. C. Tolman,
G. W. Allport, Clyde Kluckhohn, H. A. Murray, R. R. Sears, R. G.
Sheldon, and S. A. Stouffer.

For me at least, this period of stocktaking resulted in clarification of
the fundamental bases of the theory of action, a tighter organization of
its various theoretical components, and an extension of its technical
relevance into areas about which I had previously had only rather
general impressions.

Perhaps the most important single result was clarification of the
relational reference of all action theory — to actor-object relations which

"Besides the chapter already cited, the following papers offer examples of the
first line of interest: parts of "Motivation of Economic Activities" (1940) and
"Propaganda and Social Control" (1942). "Age and Sex in the Social Structure
of the United States" (1942), "The Kinship System of the Contemporary United
States" (1943), and "Certain Primary Sources and Patterns of Aggression" (1947),
present instances of the second line of interest; and with the third there deal:
"The Sociology of Modern Anti-Semitism" (1942), "Propaganda and Social
Control" (1942), "Democracy and Social Structure in Pre-Nazi Germany" (1942),
"Some Sociological Aspects of the Fascist Movements" (1942), "The Problem
of Controlled Institutional Change" (1945). My Essays in Sociological Theory
[23] includes most of these papers and a complete bibliography to 1953.

Psychological Theory in Terms of Theory of Action 623

could not be abstracted from the relationship and ascribed to one or the
other relatum apart from it. The central application was to the concept
of value — as concerned with the relation of actor and object. For ex-
ample, Max Weber had placed values in the actor, as "subjective" in
that sense (I had tended to follow him in this) ; whereas W. I. Thomas
placed values in the object (as in his well-known distinction between
attitudes and values). Neither view seems satisfactory. Once values are
treated as relational, however, belonging neither in actor nor object, but
characterizing the relation between them, then making values the focus
of the organization of systems of action becomes immediately feasible.
Along this path, a fundamental solution of the problems of the nature
of internalization and institutionalization and their relations to each
other was made possible.

A second important result, was establishing a clear logical relation-
ship among three fundamental reference points for the analysis of
systems, namely, personality, social system, and culture. This was done
by showing how they could all be systematically derived from the basic
frame of reference of action. Only much more recently have I begun
systematically to relate the other systems of action to the organism.

Finally, the most fundamental extension was into the field of c 'be-
havior psychology" on the level of animal learning and elsewhere. For
clarification of the starting points for this extension I have above all
Tolman, but also Sears, to thank, and for a great deal of follow-up —
much of which will be included in this essay — James Olds. The
reductionist trend of much behaviorist psychology,, particularly perhaps
of Watson and Hull troubled me. I did not see how behaviorist theory
could be so adapted as to recognize that the theoretical contributions of
sociology and of personality psychology on the Freudian level dealt with
more than epiphenomena.12

This more general stocktaking proved also to be the occasion for a
further reconsideration on my own part of the status of the sociological
branch of the theory of action. This resulted in the publication of The
Social System almost simultaneously with Toward a General Theory of
Action [22, 26].

The broad outline of general theory documented in those two
publications was still not complete, however; there occurred further in-
ternal developments in the main structure which were documented in
the Working Papers in the Theory of Action [24], written in collaboration
with R. F. Bales and E. A. Shils. This represents a position on the more
general theoretical levels which has remained essentially stable and pro-

32 This is one reason why at an earlier period Tolman and Kohler impressed
me. They certified that a type of psychological theory different from that of Hull
and Watson could be scientifically respectable.

624 TALGOTT PARSONS

vides the main outline for the present exposition.13 Intimate collaboration
with Bales and digestion of the implications of his research on small
groups was perhaps the most important new influence operating in the
intervening period.

In the years since the Working Papers appeared, the most important
theoretical work has been "application" and refinement through codifica-
tion of previously available materials and extension of theory into two
principal, and widely different, fields. The first of these is the relation
between family structure and the socialization of the child as illuminated
by the sociology of the family, the analysis of small groups, comparative
kinship, Freud's theory of psychosexual stages, and the psychology of learn-
ing. This has been documented in Family, Socialization and Interaction
Process with Bales, Olds, Zelditch, and Slater [25]. The second is a
reconsideration of the status of economic theory in its relations to
sociology. This study has shown that economic theory is a special case
of the general theory of social systems, and hence of the general theory
of action, documented in Economy and Society [27]. Thus, the same
conceptual scheme has proved to organize available facts and empirical
generalizations on a detailed level in such divergent fields as the
socialization of the child and the functioning of the modern industrial
economy. This result increases confidence that the theory of action does,
in fact, possess a high level of both generality and power in the analysis
of empirical materials. We have made tentative beginnings of a similar
exploration of the relations of political theory to the general theory of
action, but the results are not yet ready for publication.

Orienting attitudes. One of my most important intellectual impres-
sions was derived from A. N. Whitehead's conception of science, par-
ticularly as stated in his Science and the Modern World [37]. Three
points stand out : first, his strong emphasis on the importance of systematic
theory and the special power inherent in a well-integrated theoretical
system; second, his views of the nature of the abstraction involved in
scientific theory, particularly as related to what he called the "fallacy of
misplaced concreteness"; third, his view of the continuity of the whole
empirical world including both physical and social-behavioral areas.
Thus his use of the concept "society" to refer to phenomena of atomic
physics seemed to me more than merely metaphorical. Certain "organic"
or in some sense "antiatomistic" features of his views on all these levels
have appealed to me. I have never been attracted by theories which have
tried to build up behavior systems out of discrete isolated conditioned
reflexes alone, or social systems out of discrete isolated "individuals"

13 Chap. 3 (with Bales), "The Dimensions of Action Space" and Chap. 5 (with
Bales and Shils), "Phase Movement in Relation to Motivation, Symbol, Formation,
and Role Structure."

Psychological Theory in Terms of Theory of Action 625

alone. The Whiteheadian views of the importance of relational inter-
connectedness in systems, of organization, have appealed to me pro-
foundly.

Others who have helped to crystallize my views of science and the
role of theory in it are L. J. Henderson, James B. Conant, W. B. Cannon,
Vilfredo Pareto, and Max Weber. Conant's general views of the role
of theory in science are especially important, notably his use of "reduc-
tion in the degree of empiricism" [cf. 5, chap. 1] as a fundamental
criterion of scientific advance. Also his examples from the history of
science which have shown (as in the case of Galileo and the problem of
the limitations of the height to which a column of water could be raised
by a suction pump) that knowledge even of all the critical facts is not
sufficient to ensure a "right" or maximally fruitful theoretical explanation
of an important phenomenon [cf. 6]. In Cannon's case it was particularly
his conception of physiological equilibrium as the homeostasis of a
boundary maintaining system which provided the important model
[cf. 7].

Both as interpreter of Pareto and in his own right, Henderson was a
most important influence with respect to the concept of system and its
importance to science, and also to related concepts like equilibrium [cf.
12]. It was through the Henderson-Pareto influence that my conception
of social system in a fully technical sense first crystallized. Schumpeter
played a similar role with respect to the idea of system in economics.
Weber was, in this area, a more diffuse influence, above all in showing
the possibilities of strict scientific methodology for dealing with "human-
istic" and historical-cultural materials. His ideas of Verstehen [cf. 36]
helped very much to break the monopolistic claims to scientific standing
by behaviorists of the extreme school, who would not grant that data
concerning anything but bodily movements could be "objectively55
studied.

My general orienting attitudes toward social science, then, have
come to center about the problem of the nature and role of systematic
theory in this field. Early biological interests, reinforced by later concern
with problems of medical practice, gave me a strong conviction of the
fundamental continuity between the organic world as studied in the
biological sciences on the one hand, and the world of human social and
culturally oriented behavior on the other. At the same time, I could
not accept the kind of "reductionist53 view which maintained that the
"real53 determinants of all human behavior were to be found in the
structure and physiological processes of the organism as treated by earl}'
twentieth century biological science — with the implication that the
concerns of sociology, economics, etc. were with purely epiphenomenal
manifestations of these "real" factors. This is to say that I was deeply

626 TALGOTT PARSONS

involved in the "war of independence" of the social sciences vis-a-vis the
biological. The basic difficulty has been resolved by attempting to place
these sociocultural concerns in the context of an evolutionary view in
which they represented levels of organization of the processes of life
rooted in, but emergent from and to a degree independent of, those
which have been the more conventional biological concerns.14

Closely linked with this set of attitudes have been those concerning
the relation of theory to empirical observation and concerning the nature
of the systems dealt with in this area. With Whitehead, Conant, and
others I have had a strong conviction of the independent significance
of theory; I have never been sympathetic to a view of the methodology
of science which gave overwhelming emphasis to empiricism and in-
duction and made legitimate theory no more than a set of statements of
validated empirical fact, arrived at without benefit of theory. Opposed
to this I have set for many years Henderson's well-known definition of
fact as "a statement . . . in terms of a conceptual scheme" [11]. This
is not in the least to derogate the importance of facts but rather to chal-
lenge the claim that knowledge of fact has a near-monopoly of scientific
importance.

With respect to the nature of systems my essential "prejudice" has
been against the common "elementarist" position. This has tended to
hold that, if elementary units can be isolated and studied in sufficient
detail, then the processes of complex systems built up of such units will
become understandable without further ado. In the psychological field,
perhaps the most prominent issue has been whether the stimulus-re-
sponse unit or the conditioned reflex could be made the basis of a com-
plete understanding of psychological systems, so that independent analysis
of personality and its subsystems would become unnecessary. Similarly,
as a sociologist, I have been sensitive to the common claim that only
through understanding "the individual" independent of his social rela-
tionships could the understanding of social systems be approached be-
cause, after all, "society is composed of individuals." Here the essential
point is that organization in the sense outlined earlier, must be treated
as an independent factor in the functioning of systems, biological or be-
havioral, a factor which is not reducible to the properties of separately
given units.

All these basic orienting attitudes have applied to the general field
of analysis of human behavior, without special reference to its psy-
chological aspects. They constitute, however, the framework within
which I have approached psychological theory. Having, as a sociologist,
been deeply engaged in the battle for the independence of social science

14 The approach arrived at from this perspective seems to have converged
notably with the development of biological theory itself within the last generation.

Psychological Theory in Terms of Theory of Action 627

from biological science, I could not be attracted by a type of psychology
which was simply a branch of human biology in the traditional (early
twentieth century) sense, nor by one which dealt with psychology as the
"science of behavior" of the individual without reference to the in-
dependent significance of social and cultural systems. My experience
with such problems as that of motivation in occupational roles and in
psychotherapy convinced me that psychological help was needed to at-
tack many empirical sociological problems. To be helpful, however, it
had to be a psychology which could fit with the analysis of social systems.
For example, it had to be a psychology able to recognize that the dif-
ference between the behavior patterns institutionalized for the role of
physician and the role of business man was more than simply a matter
of the way different types of personalities happened to behave.

From this attempt to fit psychological theory into the requirements of
social system and cultural theory and to give it a place between social
systems and the organism I have derived the principal points of reference
for defining the significance of psychological theory and for specifying
the kind of theory which could acquire that significance. The con-
viction that such theory could form an integral part of the more general
theory of action, which is the main guiding line of this essay, grew up
only gradually and has not become fully crystallized until quite recently.
That this should be so, however, seems to me to be a reasonable inference
from the results of various previous attempts to codify the relations of
sociology and the psychology of personality, and it seems also to fit well
with the general conception of the nature of scientific theory which I
have put forward.

It seems best, with this discussion, to pass immediately to the out-
line of the main theoretical system itself. Such general methodological
issues as the problem of prediction, of the role of models, of quantifica-
tion, and of the formal organization of theory can be more profitably dis-
cussed when the main outline of the scheme is before the reader.

Part II

THE GENERAL THEORY OF ACTION AND ITS APPLICATION
TO PSYCHOLOGICAL SYSTEMS

Structure of the theoretical system. The structure of the theory of
action as a system will be initially discussed in two parts: ( 1 ) the "frame
of reference" or set of postulates involved, and (2) the principal
properties of empirical systems and units in systems which are made use
of in the theory, including parametric "givens." The classification of

628 TALGOTT PARSONS

variables as independent, dependent, and intervening will be discussed
under (2).

1. The theory deals with systems composed of units (in the sense
in which particles or cells are units) ; there is no system without at least
two units. What for a given analytical purpose is treated as a unit of a
system may, at the next more microscopic level of analysis, be treated as
itself a system. When an entity is treated as a unit, its properties are
always imputed to the unit as a whole and their "sources" internal to the
unit are not identified; the properties are qualities or performances of the
unit as such. When it is treated as a system, the attention is focused on
the internal processes by which given qualities and performances of the
system become understandable. Every system in turn is potentially a unit
in some more macroscopic system.

Units of systems of action are both "actors" and "social objects"
according to the point of reference. A unit is an actor when it is con-
ceived as "orienting to" one or more other actors and performing or
"overtly" acting in terms of its orientation; it is a social object when
conceived as being oriented to and acted toward by one or more other
actors. The same concrete unit may, of course, be both actor and object.

From the point of reference of any given actor, all objects which have
meaning to it are "situation." The situation as differentiated into con-
cretely discriminable entities is composed of objects (as distinguished
from abstract conceptual entities like colors or shapes). Social objects
are objects which are also actors, i.e., action systems of persons or col-
lectivities, which therefore can be treated as interacting with the actor
of reference, ego. Other objects are (a) physical objects, which have
physical spatio-temporal existence and various types of meaning to
actors, but are not treated as interacting with ego in the technical sense,
and (b] cultural objects, namely, patterns of meaning which can be
learned and otherwise oriented to (e.g., a proposition can be "believed"
or "disbelieved"), but are not treated as interacting with ego (e.g., the
proposition does not seek to "convince" ego, but only some other actor
who believes it).

Any entity which constitutes a meaningful unit in a system of action
or in its situation may be treated as an object, or if conceived as "acting"
meaningfully, as an actor. Individual human beings of course are actors,
though very generally for the purposes of the analysis of many social sys-
tems, it is the sector of the personality involved in a role, not the total
personality, which is the significant unit. Not only individual personalities
and subsystems of them but collectivities may be treated as actors. Simi-
larly, in the other direction, units or subsystems of the personality may
be treated as actors (e.g., the ego or superego) and also such organic
subsystems as, e.g., Olds' s cell assemblies [20, pp. 107ff.].

Psychological Theory in Terms of Theory of Action 629

What from the viewpoint of any given actor-unit is situation is 'di-
vided as follows :

1. The qualities and performances of other actor-units in the same
system of reference.

2. Qualities and performances of units of cognate systems with which
ego's membership system (or he himself in his membership capacity or
role) interacts.

3. Objects belonging on a system level of lower order in action terms
than the system of reference. Physical objects belong in this category,
for as we use it, the concept is not an ontological one; it is relative to
system-reference in the theory of action.

4. Objects belonging on a system level higher in the order of system-
reference than that of reference.

Some collectivities as objects, and some cultural objects at least, be-
long in this category; 2, 3, and 4 are situation to the system chosen as a
point of reference.

Actors are "oriented" to objects in their situation in so far as the ob-
ject (or a category of objects) in its relations to him may be said to
have acquired a pattern of meaning to the actor in question which is
relatively stabilized and can therefore serve as a reference point for
analysis of his action. Meaning is, in the most elementary terms, resolv-
able into two components: (a) "cathectic" meaning, as a goal object
(or object to be avoided) or source of gratification (or deprivation),
(b) "cognitive" meaning, as part of a relatively stable "definition of
the situation." Instrumental or means objects have primarily (though not
exclusively) cognitive meanings to the actor. Gathectic meaning answers
the question of degrees and kinds of wanting or not wanting to stand in
a given relation to the object; cognitive meaning the question of what
the object is in a sense significant to action, but independent of ego's
cathectic relation to it including what it can be "used for." A value, or
an evaluative meaning, is an organized pattern of both cognitive and
cathectic components which can be used to formulate a relatively stable
general orientation of an actor or class of actors to an object or class
of objects in the light of its relation to partially equivalent alternatives.

Interaction operates in the first instance through communication.
Communication is a type of act involving the transmission of meanings
common both to the agent and to the recipient object. On the part of
the agent the meaning is "intended" (not necessarily consciously) and on
the part of the recipient, is "understood" (again not necessarily con-
sciously). All communication operates through signs or symbols, acts, or
situational consequences of antecedent acts (e.g., artifacts), which can
have intended meaning to the agent and can be "understood" by the
recipient of the communication.

630 TALGOTT PARSONS

All communication, hence all interaction, implies "normative con-
trol" of action. Only by the observance of conventions or rules regarding
the "proper" meanings of signs and symbols is effective communication
possible. However arbitrary the sign may be, from an intrinsic point of
view, having acquired a meaning it cannot be used arbitrarily (i.e., its
meaning changed) in an interactive process without disrupting the
process.

From the theoretical point of view, action in a completely nonsocial
situation, where there is no interaction and no communication, is a spe-
cial limiting case. It is logically derived from the more general case by
suppressing certain ranges of possible variability involved in interaction,
i.e., those involved in the responses of alter to the communications of ego,
and vice versa. Only meanings originating in ego's own psychological
system, and unaffected by feedback on the communicative levels, need
be taken into account for this case.

In the most general terms, the frame of reference of action may be
regarded as a schema for analyzing mechanisms which control behavior.
Its focus is not in the first instance on the behavior processes themselves,
e.g., muscular-skeletal movements, but on the determination of when
and in what circumstances they will and will not take place, and in
what states of the actor in relation to the situation, i.e., of the system. For
this purpose, certain properties of behaving organisms must be treated
as given data, e.g., their constitutional capacities for certain types of
behavior; others, such as learned skills, can be treated as consequences
of action which in turn condition further action. There are, however,
many levels on which such data are relevant, and the theory is not onto-
logically tied to any one, but "plugs in" at any one of several. At the
"lower" limits, however, it can be said that the basic data or parameters
are the "performance capacities" of the organism (i.e., independent of
learned content) and the factors or conditions of the nonsocial situation
(which are neither artifacts nor signs). These of course include the
potentialities of both for modification, the first through "learning," the
second through "mastery."

2. Let us now turn to the characteristics of systems of action. A sys-
tem is constituted by the interaction of two or more units, empirically de-
termined at a given level in social or psychological terms and on the
microscopic-macroscopic range. We conceive a system of action to be
determined by (a) certain properties of its units and of its situation
which are given independently of processes in the system, and (fe) the
processes of the system which in turn can be subdivided into ( 1 ) proc-
esses internal to the system and (2) processes of interchange over the
boundaries of the system with its situation.

Let us start with the processes of the system itself. These (including

Psychological Theory in Terms of Theory of Action 631

both subcategories ( 1 ) and ( 2 ) ) may be described as constrained by four
fundamental sets of exigencies or "functional problems" which, taken to-
gether, are the dimensions of the space in which action processes operate
[cf. 24, chap. 3], The four exigencies to which a system of action is sub-
ject are those of "goal attainment," "adaptation," "integration," and
"pattern maintenance." These are dimensions of a space in the sense
that a state of the system or of its units' relation to each other may be
described, relative to satisfactory points of reference of course, as "far-
ther along" or less far along on each of these dimensions; a change of
state may be described in terms of increases or decreases in the values of
each of these variables. These four dimensions are conceived to be orthog-
onal; their values are independently variable in the sense that change of
state with respect to any one cannot be interpreted to have an auto-
matically given relation to change of state in any of the others (except
so far as this relation comes to be known and formulated as a law of
the system). It is also true that maximization of all four, and probably
of any two, is not possible in the same state of a given system.

As an essential point of reference for defining the four functional
exigencies or dimensions of systems, we assume one law, or postulate,
according to the way it is viewed. This, we call a law of inertia, on the
analogy (or more than that) of the use of the term in classical mechanics.
The law may be stated as the proposition that a process of action (as
part of a system of action) will tend to continue with its direction and
potency (see below) unchanged unless it is deflected or otherwise
changed by the impingement of some other process (in the system or in
its situation).

Very closely related to the concept of inertia is that of equilibrium.
Indeed the latter may be regarded as a special case of the former, where
a system, rather than one of its units, is taken as the point of reference.
Equilibrium is the principle that a system will tend to remain in a given
state (including stability in the operation of processes — it most emphati-
cally does not imply a state where "nothing happens") unless and until
it is disturbed by some influence from outside the system. Furthermore,
if such a disturbance occurs, tendencies will be set up to bring about the
state in which the system would have been had the disturbance not
occurred (this formulation allows for the state of a system to be defined
as conformity with a pattern or trend of orderly change, e.g., the growth
curve of weight of a child) .

The degree of stability of a state of equilibrium is of course an em-
pirical question. Some equilibria are highly stable, i.e., the forces tending
to maintain or restore the initial state are very strong; others may be
highly unstable, i.e., a relatively slight disturbance may precipitate funda-
mental changes which make restoration of the original state altogether

632

TALCOTT PARSONS

impossible (e.g., detonation of nitroglycerin by an electric spark produces
a violent change; in certain respects the nitroglycerin is in a state of
unstable equilibrium). Still other equilibria fall between high stability
and high instability.

We define the "tendency to seek goals" not in terms of any specific
propensity of organism or personality or social system, but in terms of the
concepts of inertia and equilibrium as applied to a system. From the
concept of cathectic orientation it follows that an actor-unit or system
will develop differential evaluations of different objects, and of different
relations to the same object (or category of objects), in its situation
in different circumstances. Once a pattern of such orientation has
become established, there will be an optimum relation to a given
object, an approximation to which we may call the "consumma-
tory'3 or maximum-gratification state.15 If both the state of the system
(or systems) of which the actor is a part and of the relevant situation
could be assumed to remain stable, the principles of inertia and of equi-
librium would tell us that the tendency with respect to any given ob-
ject-relation would be for it to remain in the optimum consummatory
state (this abstracts from the possibility of changing orientation patterns
by learning) .

For most empirical systems, this is a radically unrealistic assumption.
States of the system (i.e., the relations between the unit of reference and
other units) and states of the situation are continually changing. Such
changes will bring about discrepancies between the actual (and over cer-
tain periods expected) states and the optimum consummatory state. From
the concepts of inertia and of equilibrium, therefore, we derive the tend-
ency to change the state of the system and its relation to the situation in
the direction of a closer approximation to the consummatory state. This
is what is meant by the tendency or drive to attain goal states. We treat
it as a property of all systems of action, physiological, psychological,
social, and cultural, resulting from the consequences of disturbance in
the optimum relations between system and situational object.16 The same
general theoretical reasoning applies to the other three functional prob-
lems of the system.

If we assumed as a limiting case a system of action in a situation con-
sisting only of one un differentiated significant object, it would be im-

35 Such a state may, in certain cases, be defined in terms of rates of inputs and
outputs to the object; hence it is not a "static" state of relation.

16 In discussing goal attainment, it is particularly important to keep system
references clear. It is a category of the relation between a given system and its
situation. It is particularly dangerous to jump from the goal of a unit of a system
to the situation of the system (rather than of the unit) since the relation of the
system to its situation is never a simple function of the properties or state of one
of its units in relation to the situation. Situation for the unit consists primarily in
other units of the same system.

Psychological Theory in Terms of Theory of Action 633

possible to discriminate between the problem of goal attainment and
that of adaptation.17 But most systems of action function in situations
differentiated into a plurality of significant objects which present differ-
ent conditions of goal attainment. To some degree, courses of action
oriented to one goal are incompatible with those oriented to another. Yet
some courses of instrumental action, e.g., the acquisition of facilities, can
serve the attainment of a plurality of goals, and the decision among the
goals need not be made until a relatively late stage in the sequence. A
type case is the earning of money in a modern society; activities devoted
to the acquisition of money resources need not involve firm and specific
advance commitments as to the exact disposal of the proceeds for final
consumption.

As distinguished from goal attainment, adaptation is the degree to
which a system has developed a generalized capacity to meet the exigen-
cies imposed by an unstable and varying situation, without reference to
any one particular goal interest. When the system of reference is the
total society as a social system, this is its economic level of income or
wealth. Since goal attainment and adaptation are independent,18 on a
level where plural goals are involved, it is possible, and to some degree
inevitable, for them to conflict. For the generalization of facilities rele-
vant to a wide range of goal interests may be bought at the cost of loss
of particularized suitability for any given specific goal. Reconciling these
two bases of interest is one primary focus of the integrative problem in
systems of action.

The two functional exigencies of action systems so far discussed, goal
attainment and adaptation, concern relations between the system and
situations external to it. The other two concern problems internal to the
system, i.e., conditions of its stability which, in the analytical sense, are
independent of situational conditions. We have called these two internal
exigencies, pattern maintenance and integration respectively.

As we define it, a system of action is a system of relations between
living organisms and objects in the environment. It is only a system of
action in the technical sense so far as the relations are organized through
learned patterns of orientation.19 Once thoroughly learned (in a person-
ality sense, internalized) such patterns become the primary focus of the

17 This case is presumably approximated in the "mother-child identity" of
infancy [cf. 25, Chap. 2].

18 But, of course, also interdependent. We assume them to be orthogonal
dimensions.

Jtt The case where the environmental event is only a "trigger release" for a
constitutionally built-in pattern of behavior is, in action terms, a limiting case
[cf. 25, Chap. 4]. The crucial point is the significance of the consequences of re-
sponse as a basis of learning, and hence modification of subsequent behavior. If
the consequences have no effect on subsequent behavior, the phenomena are of no
theoretical interest from our point of view.

634 TALCOTT PARSONS

organization of the system as a system of action. It is in this sense of
organization on a pattern, i.e., "cultural," level that we have introduced
the concept of value.

There are two primary aspects of the functional problem of pattern
maintenance. The pattern system (which on a human level is certainly
"cultural" whatever term may be used for subhuman analogues) which
regulates any particular system of action is always part of a larger system
of patterns, related to other elements in terms of "meaning-congruence,"
consistency, etc. In the limiting case of a total society, the total meaning
system may be independent of any others, but it is still subject to the
exigencies of its "making sense" to the actors in the system.20 In any case,
some part of the relevant pattern may be threatened by actual or ap-
parent incompatibility with other parts of a larger system.21

The type of potential disturbance with which we are at present con-
cerned may arise from outside the system in question through posing
questions of compatibility in inescapable form. It may also arise from
within, through weakening the specific mechanisms ("defenses")
through which "intrinsically" incompatible elements are enabled to co-
exist in the same system. Indeed, in a slightly different perspective, what
we refer to here as the function of pattern maintenance might be called
the "strain toward consistency" in the pattern system. This is the active
version which is parallel to "goal-striving."

The second primary aspect of the functional problem concerns not
the consistency of the pattern system itself, but the level of motivational
commitment to implementation of the pattern or of some subpattern
component of it. It is for example possible to "believe" a proposition in
the sense of assenting to its truth when involuntarily confronted with a
situation where it is impossible to evade taking some position; it is
quite another thing to believe it as a focal center of primary orientations.

20 This would be strictly true only in a limiting case. Historical tradition and
the presence of other societies "frame" the meaning-problem of a particular
society in reference terms wider than the psychological needs of its members.

21 We assume here the psychological validity of the "principle of contradiction,"
namely, that it is not possible, without strain and the operation of specific
mechanisms, to hold two or more mutually contradictory beliefs at the same time,
e.g., that Boston is northeast of New York and that Boston is southwest of New
York. This constitutes a fundamental reference point for psychological (and socio-
logical) as well as logical analysis. Further, we hold that in the cathectic-ex-
pressive field there is a similar principle of congruence according to which it is not
possible without strain and the operation of specific mechanisms to be committed
or attached to two or more mutually incompatible cathectic commitments at the
same time. For example relatively total "love" and "hatred" of the same person
is only possible if mechanisms of defense prevent the full juxtaposition of these at-
titudes. This is why the cruder ambivalences which are not reinforced by other
strategic factors cannot survive good psychotherapy.

Psychological Theory in Terms of Theory of Action 635

This problem of motivational commitment is what we call the "tension-
management" aspect of the pattern-maintenance function. (Tension is
here used as a general psychological term in motivation theory for a
state of unstable equilibrium.) So far as there is tension, there is a "dis-
position" toward change of state. This may potentially take a direction
which will lead away from conformity with the cultural pattern in
question.

The generalized significance of this problem derives from the inter-
penetration of psychological, social, and cultural systems. Broadly, we
may say no internalized cultural pattern system is ever entirely idio-
syncratic to the particular personality. Short of the total personality, any
partial psychological system must reckon with the repercussions of the
rest of the motivational system on this particular partial system. Con-
versely, no human cultural system ever "operates" without being borne
by a multipersonal social system and the several participating person-
alities. The system exigencies of cultural consistency and the operative
actions of both social and psychological systems always impinge on the
tension problem. In other words, cases of relative stability of pattern
maintenance must always be accounted for by specific mechanisms
which "forestall" the continually arising threats to this stability. It is
never safe to assume that a cultural pattern is "naturally appropriate"
and will be maintained in a system of action just because it "has to be
that way."22

The two primary aspects of the pattern-maintenance problem be-
long together because their outcomes flow into the same channel, namely,
either reinforcement of the conformity of action with the values and
expectations defined in the cultural pattern system, or reinforcement of
tendencies to deviance from these expectations. The pattern system is
the fundamental point of reference for analyzing the problems of stability
and instability of systems of action.23

The pattern-maintenance function refers to the state of the unit, and
the conditions of its stable equilibrium which are relatively independent
of its position as a unit in this particular system of action. Essentially we
may say that the foci of these changes are (a) the "culture" relatively
independent of specific action-system involvements and (b) the "person-
ality," in the sense of the impingement of the motivational system as a

22 In terms of the hierarchy of controls discussed above the consistency aspect
of the pattern-maintenance problem looks "upward" to the central source of
control, the "tension" aspect looks "downward" to the units subject to control.

23 We may suggest that stability-instability is the best pair of terms to use for
the system level of reference. They refer to the concept of equilibrium, and
through it, to inertia. On the other hand, the terms conformity-deviance are best
used with reference to the unit level. A unit conforms to the norms of the system
or does not; but a system is stable or unstable.

636 TALGOTT PARSONS

whole on the particular action system in question. This is indeed the
center of the problem area which has been called "culture and person-
ality."

We still have a fourth functional problem of the system of action
which, from the point of view we have discussed, must be treated as
primarily "internal" to the system but at the same time must be distin-
guished from the pattern-maintenance function. This is the function we
have called the integrative. From the point of view of the unit in a sys-
tem, a culture pattern has in the first instance cognitive significance ; it is
content, in the sense of information, to be learned in the double sense of
comprehension and motivational commitment. After being learned it
becomes a property of the unit itself. But the units of a system are also
objects to each other in a predominantly cathectic sense. By system
integration we mean the mutual cathectic adjustment of these units to
each other in the perspective of the internal harmony or, as is often said
for social systems, solidarity or cohesion of the system. Every system then
has a level of integration which is a function of the "object-relations" of
its units to each other, of the adjustment of their mutual cathexes
through motivational mechanisms. If the units are persons or their roles,
this takes place through what have been called the mechanisms of ad-
justment [cf. 25]. If the system is intrapersonal, it is through the mech-
anisms of defense. Mutual antagonism or aggression (intrapersonally,
"conflict" ) is of course prima facie a threat to integration. System inte-
gration and pattern maintenance are dynamically interdependent, but
much analytical and codifying work makes it clear that it is essential to
discriminate them as independent variables.

These four are the fundamental variables of our system. Before dis-
cussing some of the parametric24 categories which are essential to give
the system empirical determinateness (at different levels), one funda-
mental relation among the four variables must be mentioned. It has
already been noted that two of them refer to aspects of the state of the
system in relation to the situation external to it, whereas the other two
refer to aspects of the internal organization of the system. Let us eluci-
date some implications of this differentiation.

A system of action is, we have held, a "boundary-maintaining" sys-
tem. There must be, then, in the relevant respects, a closer order of inte-
gration or organization within the system than between the system and
other systems. We take this to imply that units operate with a greater
scope of freedom or autonomy when they are functioning in intersystem

24 By the terms parameter and parametric I refer to propositions which are
empirically essential to determinate analysis by use of the theoretical categories
of the system but not values of these theoretical variables as such. A parameter
states given data for an empirical problem. The frame of reference of a parametric
statement must, of course, be congruent with that of the theoretical system.

Psychological Theory in Terms of Theory of Action 637

relations than when they function in intrasystem relations. In so far as
the hierarchy of systems of action is a hierarchy of levels of control of
behavior, there is a difference of level between goal attainment and
adaptive references on the one hand and pattern maintenance and inte-
grative references on the other. The former pair look "downward" in the
scale toward the remoter situational factors which are relatively independ-
ent of the organizational system of reference. The latter pair look "up-
ward" toward the more central foci of the total system of control.25

This distinction is not ontological but entirely relative to the place of
the given system in a larger reference framework. Such a larger reference
framework is inherent in the general conceptual scheme of action; the
relative treatment given on the one hand to goal attainment and adaptive
problems, and on the other to integrative and pattern-maintenance prob-
lems, is a function of the place of a given system in this reference scheme.
At this point, we should also comment on the relation of this four-
dimensional scheme to that of "pattern variables" which has figured
prominently in previous publications [cf. 23, part 2, chap. 1 ; 24, chaps.
3, 5]. The essential point is that the four dimensions incorporate the
core of the pattern-variable scheme. The difference is that the latter
makes explicit the basic distinction between the "attitudinal" and the
"object-categorization" aspects of the general action frame of reference.
That is to say the attitudinal and object categorization subpairs of pat-
tern variables can be "matched" so that functional specificity and uni-
versalism become the two relevant aspects of the adaptive dimension.,
affectivity and performance, those of the goal-attainment dimension,
functional diffuseness and particularism of the integrative dimension,
and affective neutrality and quality of the pattern-maintenance dimen-
sion. Of the original five pattern variables this omits self vs. collectivity
orientation. This last, it has become clear, is a category referring to the
relations between systems of action, not to the constitution of any one
particular system-
There has been criticism of our formulation of the pattern-variable
as dichotomies: affectivity vs. affective neutrality, universalism vs. par-
ticularism. It now seems to be clear that they are dichotomous because
of the location of their reference to the integrative problem within sys-
tems of action. Interests in goal attainment stand in an inherent poten-

25 When generalized, this distinction can serve as a principle for relating systems
to each other in a hierarchical series. Put otherwise, a hierarchy of control, such
as we have discussed, may be conceived as a series of alternating layers of adaptive
goal-seeking components and integrative pattern-maintaining components. Accord-
ing to which pair has the functional priority we can then distinguish two types
of system — one more situation- and performance-oriented, the other more internally
and "expressively" oriented. This distinction will be used later in connection with
psychological systems.

638 TALGOTT PARSONS

tial conflict with those in pattern maintenance because the former can
always easily be dominated by questions of "expediency," the practical
(i.e., situational) availability of objects of consummatory gratification.
Similarly, the opposition between adaptation and integration derives
from the dilemma between the unit's orientation to the external "fence-
mending" functions and its integrative adjustments to other units in the
system. Because adaptation is the generalized focus of situational rela-
tions, it is the focus of conflict with integrative interests.

Our most central methodological distinction is that between funda-
mental theoretical variables of our system and parametric categories. The
latter can be divided into two classes, those characterizing the units of
the system, and those characterizing the situation in which the system
functions. In common with other fields of science we distinguish, with
respect to both, relatively stable properties of objects, and time relations
[cf. 24, chap. 5].26

Time relations present what, for us, is the simpler problem. All the
empirical sciences take it for inexorable fact that certain events have
occurred at given times and in given time sequence. Given certain ante-
cedent time determinations other time determinations can be deduced
by theoretical reasoning; this is what we mean by prediction in its tem-
poral aspect. But time is never a manipulable variable; time is a frame
of reference within which one can state and interpret the assumptions
about and the consequences of the operations of manipulable variables.
When we say that we decide "when" something will be done, we do not
manipulate time but the variables which have consequences in time. Like
all parametric features of systems, temporal relations have two orders
of scientific significance: (a) time is one fundamental aspect of the
givenness of the empirical world which provides the empirical base from
which any deduction or prediction can be carried out; (b) in the sense
of when a given future event will occur relative to others, time is a
fundamental aspect of the empirical manifold to which any chain of scien-
tific reasoning will lead. If time (possibly period rather than instant is the
relevant unit) cannot be specified, empirical determination is incomplete.

For purposes of the theory of action, the properties of objects, both
of units of systems and of situational objects other than time order, can
be reduced to two classes: their value orientations, and their "potency."
Value orientation is location in a system of reference formulated in terms
of the four dimensions we have discussed. This aspect will be further dis-
cussed when we take up the internal differentiation of systems of action.
Here, let it be noted only that location can be viewed "statically," in terms
of place in a structure, or "dynamically" in terms of processes of change of
location; or the two can be combined in the concept of "orbit" [cf. 24,

26 For the view that time relations constitute, for physics, a set of parameters,
not of fundamental variables, cf. [3].

Psychological Theory in Terms of Theory of Action 639

chap. 5]. Theoretically these are different derivatives from the same
theoretical analysis, and the distinctions do not raise any special problems.

Potency is a new term we have introduced in this essay because we
have not heretofore had a single term to cover what we mean for physio-
logical, psychological, social, and cultural systems. By potency we mean
relative degree of importance as between units in affecting the outcome
of processes in changing states of the system. It is the analogue of mass
in classical mechanics. For psychological systems the best formulation
available to us seems to be that of "motive force" as used by Olds [cf.
20, pp. 11 Off.]. For social systems, with some qualifications, prestige
seems to be the best available term. Looked at in performance terms,
potency thus is relative capacity to influence the outcome of a process.
The rank order of units with respect to this capacity is, so far as it is
legitimized by values, the stratification of the system. It is a function of
integration with the value system, internalization or institutionalization,
but not only of this.

For purposes of a given analytical procedure, both values and po-
tency, like temporal relations, are given either as among the data of the
problem or as empirical resultants of the process being analyzed. This
holds so far as strict analytical procedure in terms of a given system refer-
ence is concerned. Intrinsically, of course, all these data are subject to
change; only for specific methodological purposes can they be treated as
given.

Here the distinction between the properties of the units and the
properties of situational objects becomes essential. A change in the major
properties of units and/or of their patterns of temporal relationship is
to be regarded as a parametric change in the state of the system, not as
merely an "equilibrating" process. In psychology, the distinction is that
between learning processes and performance processes, in social system
terms, between "social change" and "normal functioning."

To analyze processes of learning and of social change theoretically,
it is essential to take account of multiple system references. The unit can-
not be treated as the object of an explanatory problem hi other than
"positional" terms unless it is itself treated as a system; hence its relations
to other units are not intrasystem relations but these others are treated
as its situation. This distinction between "positional" change (including
both direction and rate) and parametric change is fundamental to the
theory of action, and we believe to all other comparable theoretical
schemes.

We believe that classification of variables as independent, dependent,
and intervening cuts directly across the classification built on the distinc-
tion between fundamental system variables and parametric categories.
The independent-dependent-intervening variable scheme refers to the
logical operations involved in an empirical problem-solving sequence.

640 TALGOTT PARSONS

The system-variable-parameter scheme refers to the logical structure of a
theoretical system, which is never relevant only to one problem orienta-
tion, or totally involved in the same way in different ones.

In a given case, the values of one or more of what we have called the
fundamental system variables or of significant parametric categories may
be those which serve as systematic or empirical independent variables.
Then the values of one or more other fundamental system variables or of
one or more other parametric categories may be the corresponding de-
pendent variables. The intervening variables would this time be the
values again of one or more parametric categories which were not di-
rectly observed but inferred from the data on the values of independent
and dependent variables. This statement implies that the laws connecting
the values of the system variables are known. So far as this is not the
case, one or more of their values might serve as an intervening variable.

Although there seems to be no very specific rule, it appears that the
most general case of system analysis is what we would call input-output
analysis. In the present methodological terms, this would involve treat-
ment of a significant situational parameter as the independent variable
and usually one or more other situational parameters as the dependent
variables. A case would be the introduction of deprivational changes in
the situation which would reduce the input of goal gratification into the
system. The analytical procedure would trace the repercussions of this
input change through the operations of the fundamental system vari-
ables and thereby describe a new state of equilibrium of the system with
altered outputs from those obtaining in the initial state. Both the output
values, which would usually be empirically ascertained through the
states of the situational parameters, and the values of the system variables
would change, and any of these could be the empirically determined de-
pendent variables. (See under Construction of function forms below.)

Before we can deal with psychological systems as such, we should dis-
cuss two further general points about systems of action. The first of these
concerns the ways in which systems are differentiated and integrated
relative to their patterns of value orientation; the second concerns the
implications of the conception of boundary-maintaining equilibrium for
the problem of "normality" and "pathology."

As we have presented it, the theory provides three foci of the internal
differentiation of a system of action: (1) The units may and generally
will be differentiated in a rank order of relative potency. ( 2 ) They may
be differentiated in terms of relative position (treated statically or as an
"orbit," a range of successively occupied positions) in action space, i.e.,
of functional significance in the system. (3) They may be differentiated
with respect to rates of input-absorption and output-production in their
relations to other units.

Psychological Theory in Terms of Theory of Action 641

The differentiation of the system may follow both temporal and, in
the action sense, spatial patterns. The first we speak of as the differentia-
tion of phases of system process, the second as differentiation of the
structure of the system. The two are different aspects of the same basic
phenomena and are capable of being analyzed in terms of the same
frame of reference [cf. 24, chaps. 4, 5]. There is, however, a sense in
which phase differentiation is a more "elementary" phenomenon than
structural differentiation.

The fundamental basis of the phenomenon of phase differentiation
lies in the fact that, as discussed above, the variability of its situation
does not permit a system to remain stably in a consummatory state.
Adaptive-instrumental modifications of the system, and through these of
the situation, are necessary to optimize the possibilities of gratification
(goal attainment), and these activities have internal repercussions in the
integrative and pattern-maintenance aspects of the system. The primary
basis of phase differentiation, then, lies in the limitations on the per-
petuation of consummatory states and on the system-situation condi-
tions which lead to repetitive return of approximation to such states.
Within limits, phase differentiation is possible without structural differ-
entiation; i.e., the system can be treated as a single unit.

We know of no class of empirical systems for which it seems useful
to treat the system as one of plural units which are absolutely undiffer-
entiated except for the phases of the units; such a system is theoretically
conceivable though probably its equilibrium would be highly unstable.
The primary basis of structural differentiation is functional, i.e., in terms
of the primary "contribution" of the unit to the functioning of the sys-
tem. This primary contribution is defined as the output at the goal-
attainment boundary of the unit in question. Thus a unit differentiated
from others (i.e., specialized) in terms of adaptive function for the sys-
tem will contribute not directly to the system's goal attainment but to the
adaptive level which facilitates attainment of an indefinite number of
specific system goals.27

The goal attainment of any system (and in discussing differentiation
we must consider the unit as a subsystem) is at the same time the produc-
tion of an output to its situation (the rest of the system) and the source
of a category of inputs to itself, in this case "gratification" or reward in
some sense. Units thus will tend to be specialized with respect both to
their types of output and to their types of input, e.g., reward or gratifica-
tion source.

But, since on the requisite levels units themselves are systems, they
will tend to have not only distinctive goal outputs and rewards (as com-

2T For an analysis of the boundary interchanges between subsystems of a society
see [27, Chap, 2].

642 TALCOTT PARSONS

pared with other units) but also distinctive adaptive patterns and func-
tions, integrative patterns and functions, and distinctive subvalue systems
and pattern-maintenance functions. Not only must the goals of units be
integrated in the system, but all their other functions must also be inte-
grated.

It is clear, too, that the structural differentiation of units must be
coordinated with the phase differentiation of the system. This occurs as
different units have their goal-attainment and other phases at different
points in the phase cycle of the system as a whole. The consummately
phase in the phase cycle of a unit comes at the phase in the system cycle
where its primary function in the system has primacy for the time being.
Thus an adaptively specialized unit will come to its consummatory phase
during the adaptive (instrumental) phase of the system cycle. The con-
summatory phases of the other units then fit with other phases of the
adaptive unit, etc.

The concrete structure of systems of action cannot be derived from
this functional paradigm alone. For each unit is subjected to exigencies
other than those defined by its primary goal interest. These exigencies,
as traced through hierarchies of system-subsystem relationships, will "de-
flect" the structural patterning of the system in certain respects from the
"pure type" of a functionally differentiated system. We hypothesize that
of the three functional needs other than the goal interest in general the
value pattern of the unit will be least deflected by varying exigencies (if
the unit arises by differentiation; where, as in social systems is often the
case, it has recently "joined" the system the deflection may of course be
considerable). The second order of deflecting exigency will be the inte-
grative, and the most prominent the adaptive. Hence, next to major
alteration in a unit's opportunities for gratification, its adaptive position
is the most prolific source of change in its structure.

One further point: we have emphasized that for the strictest pur-
poses of equilibrium analysis the properties of the unit of a system of
action must be assumed to be given. This means that what we are call-
ing the structure of the system is given. Structural change then must be
treated as raising a different order of problem from that of analyzing
equilibrating processes in a system with given structure and situation.
In one major aspect, the difference lies in the fact that in the former
problem at least two levels of system relation must be taken into account.
It is no longer possible to treat the unit only as possessing stably given
properties and not as itself a system. Both the system in which the unit is
a unit, and the unit as itself a system must be treated "dynamically."
For this reason, analysis of structural change in systems of action presents
a more 'difficult theoretical problem than does analysis of equilibrating
processes of a single system of reference.

Psychological Theory in Terms of Theory of Action 643

Finally, a word about the problem of normality and pathology. We
have stressed that there is always a normative aspect in the analysis of
process in systems of action (we believe this to be the case with all phys-
iological processes as well) . This was brought out most clearly in the case
of the concept of communication ; if the conventions of the sign-meaning
system are violated, communication is disturbed. We connect this norma-
tive reference with the property of boundary maintenance which we
impute to systems of action. Boundary maintenance means maintenance
of a distinctive intrasystem pattern which is not assimilated to the pat-
terning of the extrasystem situation. In this theoretical setting, the ques-
tion of "how well" or "successfully" the system is maintaining its pattern
is unavoidable on any comprehensive level of theoretical analysis, though
for special purposes it can be avoided. Furthermore, the question cannot
be localized in any specific aspect of system functioning; it applies to the
system as a whole in relation to the situation as a whole. In a relative
sense, the functioning of systems must be evaluated: they are more or
less well adapted to their situations, more or less well integrated, etc. A
conceptual scheme which makes values a central category cannot evade
this consequence, and should not attempt to. But evaluation of the func-
tioning of a particular system in a particular situation is a very different
thing from a judgment of the value of that system and its results in some
wider frame of reference. Our theory makes no particular assumptions
on the latter level.

It may prove useful to introduce here a diagrammatic representation
of a system of action. [Adapted from 24, p. 182.]

\

/

Adaptive

Goal -attainment

phase and

phase and

subsystem

r

subsystem

"X

k

Pattern - maintenance
(and
tension -management)
phase and subsystem

J.

Integrative
phase and
subsystem

DIAGRAM 1

\

644

TALCOTT PARSONS

The letters represent the four fundamental functional system prob-
lems or dimensions of action space as follows: A adaptation, G goal
attainment, / integration, L latent pattern maintenance. Representation
of these four on a two-dimensional plane is simply a matter of con-
venience. The diagram does not represent the cross tabulation of two
variables, but each of the four sectors of the diagram represents an
independent dimension of variation. The clockwise order of juxtaposi-
tion is that involved in the phase movements of ordinary system process
(in psychological terms, performance processes) ; the counterclockwise
order is that involved in the phase movements of processes of structural
change in systems (psychologically speaking, learning). Hence the order
of juxtaposition is not arbitrary but has theoretical meaning.

We have pointed out that these four dimensions also constitute the
primary frame of reference for processes of differentiation of systems of
action. Hence the same diagram may be used to represent the primary
functional subsystems of a larger system of action and certain of their
relations to each other. Each of the subsystems will then be conceived as
engaged in exchanges of inputs and outputs with each of the others.
Seen in these terms the diagram takes the following form:

DIAGRAM 2

We will attempt to explain below the rationale of the different types
of input and output at each boundary of each of the functional sub-
systems.

Psychological systems. Having outlined the main structure of the
theory of action, the next task is to show how it can be adapted to
psychological subject matter.28 The reader should recall that in the form
in which we and our associates worked it out, the general theory of ac-
tion was not originally applied to psychological systems as such but to
social systems and then to certain of the latters' points of articulation
with cultural and with psychological systems. The application to psycho-

23 On one very important level this has already been done by Olds [20,
Chap. 4]. His treatment should be compared with the following throughout.

Psychological Theory in Terms of Theory of Action 645

logical systems thus represents an extension through codification pro-
cedures. It does not purport to discover or state new psychological knowl-
edge, but to show that certain main lines of known psychological analy-
sis can be stated and systematized in terms of the theory of action.

It is also important to keep in mind Olds's caution concerning the
level on which this attempt is made. It does not select a single funda-
mental empirical unit of all action systems and attempt to show how
different levels of system can be analyzed with it. Rather, it selects an
abstract unit, located in a particular type of space and having other
general properties, and shows that a variety of empirically different units,
e.g., Olds's "concepts/3 need-dispositions, role expectations, collectivities,
etc., can all be defined as belonging to this class of unit in an action
system. Hence the propositions general to systems of action can be ap-
plied to systems involving this type of action unit.

Let us now discuss psychological systems in general terms and then
attempt to spell out the general statement with reference to some se-
lected cases. We have defined a psychological system with reference to
behavior, i.e., a set of relations between a living organism and objects in
its environment. A psychological system is a system of behavior pertain-
ing to a particular organism. The total system of behavior of one organ-
ism is its personality, but a personality, even of a subhuman animal, is
divided into a complex set of subsystems of different types. The following
discussion applies both to personalities and to their subsystems.

Thus psychological systems stand between the organism and the ob-
ject-system presented by its environment. They are the systems generated
by the relations between these two entities. Finally it must be noted again
that a particularly crucial class of objects for behaving organisms is the
behavior, i.e., personalities, of other organisms, particularly though not
exclusively of the same species.

Like any system, certainly any system of action, a psychological sys-
tem must be analyzed in terms of two sets of processes: "boundary
processes," which involve the relations between the system and its en-
vironment, and internal processes, which involve the units of the system
in their relations to each other. Empirically these two sets of processes are
not always completely separated, but the conception of boundary main-
tenance, which is fundamental to our scheme, makes their discrimination
a primary theoretical distinction.

Not all the boundary processes in psychological systems are directly
observable. This is because one essential set of boundary relations will
be those between the psychological system and the organism; a certain
proportion of these will not be directly observable as behavior, but must
be inferred as resultants of intervening variables. The same is true of the
internal processes of the psychological system. The directly observable

646 TALGOTT PARSONS

processes then include some, but not all, the boundary relations between
psychological system and organism, and all the boundary relations to
objects external to the organism-personality system.29 Further, it must be
noted that where the psychological system in question is not a total per-
sonality but a subsystem of a personality, its important boundary rela-
tions will not be the same as those of the personality but will include re-
lations to other units of the personality system. Hence it is dangerous to
presume that the same classes of processes will be directly observable
for all classes of psychological systems.

The first substantive question we have to raise is, what are the units
of psychological systems as systems of action? In a general sense they are
components in the organization of behavior processes which have come
into existence through learning. What is learned is the meanings of ob-
jects in the situation of the organism-personality unit, and of parts and
processes of the person's own body, treated by him as objects. The units
embody what we have called elsewhere cathectic and cognitive com-
ponents organized in relation to each other in evaluative terms; the
units thus consist in organized modes of orientation to the objects in the
situation of action.

Looked at in a slightly different way, the units of a psychological
system constitute, in one respect, what is ordinarily called the needs of
the system, in another its dispositions to act, i.e., to control the capacities
of the organism and of external objects in the interest of goal-directed
behavior. Hence we have adopted the composite term need-disposition
to refer to these units [cf. 22, part 2]. It is essential to note that in our
view all of the units of a psychological system have both cathectic and
cognitive components; all of them are organized with reference to values;
all of them have both need aspects and dispositional aspects. They come
to be differentiated from each other in a variety of respects, which will
effect the relative primacies of these aspects for the different units of the
same system, but this does not invalidate the central proposition that all
aspects are present for every unit of a psychological system.

Depending on the degree of differentiation assumed with reference
to organic life in a particular frame of reference, the psychological sys-
tem may or may not be treated as "part" of the organism. If the less
differentiated frame of reference which discriminates only "organism"
and "culture (I)5' is used, then it clearly is part of the organism. If, on
the other hand, one uses the more highly differentiated frame of refer-
ence which discriminates four system types in the action field, the psycho-
logical has to be treated as an independent system level. It is on the
latter level that the present discussion will be conducted.

29 This, of course, disregards "conventional" restrictions in such observations,
e.g., where rights of privacy as between husband and wife are involved.

Psychological Theory in Terms of Theory of Action 647

In this case, a psychological system will have three principal types
of boundary relationships with other action systems (other, that is, than
other subsystems of the same personality). These will be relations with
the organism, with the social system (i.e., social objects in the situation
of the personality), and with the cultural system. In addition there will
be relations with systems outside the action frame of reference, notably
physical (including chemical) systems both "internal" to the organism
(in its "vegetative" aspect) and in the environment. Of these last no
attempt will be made here to give an account, except to note that action
systems in the organic world in general operate to control (and adapt
themselves to) the functioning of physical systems.

Within our range the relation between a psychological system and
the other three types of action system which constitute its environment or
situation may be analyzed in terms of two fundamental concepts, namely
input-output interchange, and interpenetration. Let us discuss each of
these for each of the three main boundary types.

Psychological System and Organism

It is not possible here to give a complete account of the very com-
plicated interchanges between personality and organism; only a few
highlights can be touched upon. First, on general theoretical grounds we
suggest that the inputs the personality (psychological) system receives
from the organism are in the first instance to be regarded as facilities
for its functioning; this is not exclusively the case but the facility aspect
has primacy.30

In what do these facilities consist? In accord with the paradigm
which we use to classify each aspect of the operation of a system in terms
of four functional categories, we can propose four categories of inputs.
The first, and in a sense most basic, is what is usually referred to as
motivational energy in the organic sense (which should not be identified
with motivation in a psychological sense ) . In other words the organism
is the source of the energy which underlies all processes of action. This
is the phenomenon underlying "tension" and should, in our opinion, be
treated in terms of the concept of inertia, i.e., as a flow which tends to
remain constant unless increased or decreased by special factors im-
pinging on it.

80 The general theoretical grounds derive from the fact that, as noted earlier,
in the general system of action the organism has primarily adaptive functions,
whereas the personality system has primarily goal-attaining functions. Then, on
grounds which cannot be developed here [cf. 27, Chap. 2 for the fullest exposition
yet attempted], the interchange between an A subsystem and a G subsystem is
primarily mutual interchange of facilities whereas that between a G subsystem and
an / subsystem is primarily a mutual interchange of rewards.

648 TALCOTT PARSONS

The second and third types of facility provided by the organism are
also familiar in psychological thinking. The second is perceptual (or
cognitive) capacity, i.e., the capacity to assimilate and organize "infor-
mation" coming from the environment of the psychological system, not
only from "external" objects but also from the organism through pro-
prioceptive processes. The third is "performance" or "response53 capac-
ity, the capacity to utilize the structures of the organism, notably the
skeletal-muscular structures (and through them external means-objects)
for physical manipulations of the environment.

Fourth, there are the facilities which aid in integrating these other
facilities with each other and with the needs of the psychological sys-
tem. Though much in this area is obscure, it seems correct to speak of
pleasure as the primary integrative facility. Pleasure is particularly closely
associated with the capacity to learn, constituting a proprioceptive re-
ward mechanism which can, by learning processes, be associated with
the attainment of the goals of the psychological system.

What outputs of the psychological system to the organism correspond
to these classes of inputs? Two guiding lines may be suggested for identi-
fying them. First, they should appear as mechanisms of control of the
organic processes most closely involved in behavior; second, they should
be identifiable as facilities from the point of view of the behavioral or-
ganism.

In one context the most fundamental of these psychological outputs
seems to be what Olds calls "motive force." [Cf. 20, p. 110, et passim.]
This may be thought of as that part of the energy received by the psy-
chological system from the organism which is "fed back" to motivate
instrumental processes and which can increase the performance potential
of the organism when it is controlled by the psychological system. A
second type of psychological output to the organism can be called the
"directional" component. This means that in relatively specific situa-
tions the facilities of the organism come into direct control of specific
motivational structures of the psychological system. It is the process of
immediate determination of the directions of "interest" in the perception
process, of the directions of goal-seeking in the performance process,
and forms of "acting out" for which pleasure can be a direct reward.

The third component of output may be said to be the "expectation"
component. This determines the "attitudinal set" of the organism with
reference to its integration with the psychological control system. Basically
it is the "expectation" that organic interests will be well served by "going
along" with the psychological system, i.e., satisfying psychological needs.
Disturbance in this relationship seems to be involved in the deeper types
of psychosomatic problem.

Underlying all of these is a pervasive problem of the "organic secu-

Psychological Theory in Terms of Theory of Action 649

rity" which is dependent on the stability of the whole relationship be-
tween organic and psychological systems. It is presumably on this secu-
rity above all that the stability of the organic energy flow to the psycho-
logical system is dependent

Sketchy as this account of the input-output relations between psy-
chological system and organism has been, perhaps it has been carried
far enough to suggest that a variety of problems which have played an
important part in the psychological literature can be approached in
terms of the theory of action.31

Before taking up the concept of interpenetration in more general
terms, a few preliminary remarks are necessary. At many points in con-
structing the theory of action it has become evident that analytical dis-
tinctions between types of system do not correspond to concrete systems.
A business firm, for example, may be spoken of as a collectivity with
economic primacy, as both "participating in" the economy and more
determined by its role in the economy than in any other analytically de-
fined subsystem of the society. But "the economy" cannot be defined
as the aggregate of business firms and their relations, if firms are inter-
preted to be concrete collectivities. For these units have political and
other "aspects" and many collectivities other than firms, e.g., house-
holds, have economic aspects.

Where it is necessary to speak of two or more analytically dis-
tinguishable relational systems as both constituting partial determinants
of process in a concrete empirical system, we speak of the systems as
interpenetrating. The same concrete phenomena must be interpreted as
"participating in33 both analytical systems. Clearly, in the conception set
forth here, behavioral organism and psychological system are interpene-
trating in above sense although treated as analytically distinct systems.
On the more concrete level, behavior is always behavior of the (con-
crete) organism. If the organism as an anatomical-physiological system
be reified (which, as we learn from Whitehead, is illegitimate) then, of
course, a psychological system as a distinct system cannot exist; the only
recourse is to reductionism. But from the point of view adopted here
this is not a necessary inference.

In the present context interpenetration implies that there will be

31 It is of interest in the general context of this essay to note that the analytical
model for this relation has been derived from the boundary of the economy as a
social system, the boundary which involves the input of capital as a generalized
facility for economic production, and which Smelser and I have interpreted to be a
boundary vis-a-vis what we have called the "polity," which is not to be confused
with government. [Gf. 27, Chap. 2, pp. 56-59, 70-72.] Though far less fully
developed in the psychological case, the correspondences appear sufficiently strik-
ing so that more than a mere analogy in the derogatory sense of the term seems
to exist.

650 TALCOTT PARSONS

identifiable physiological mechanisms of all processes operant in psycho-
logical and psycho-physiological systems. But processes with respect to
these mechanisms will be conceived as analyzable in terms of the inter-
action of the two system levels we have distinguished, not in terms of
either one of them taken alone. The cell assembly which Olds [20, p.
107ff.], following Hebb, postulates is a model of such a mechanism.

We believe that the stimulus-response-stimulus paradigm of the most
elementary psychological processes refers to a system in which the inter-
penetration of psychological and organic systems is treated as the salient
feature. It is significant that the cases which have occupied the greatest
attention have been those of animal behavior in highly restricted, short-
term conditions, with a specifically set goal and a severely limited range
of instrumental possibilities. These conditions altogether eliminate the
higher-order level of social object-relations, and the more complex pat-
terns of culture.

The S-R-S paradigm deals with the behavior of the organism on a
level where its mechanisms of control are relatively closely bound to
specificities of skeletal-muscular movements, so that the categories of
stimulus and response can be given quite specific physical perception and
physical movement meanings. In other words, it is truly an elementary
action system in which some components of more developed ones are
rudimentary at best, others are not differentiated from each other to a
readily analyzable degree at all.

We would expect that the closeness of fit of organic and psychologi-
cal processes would be most marked at this level. But there is no reason
to believe that the interpenetration which is so conspicuous here ceases
to exist at higher levels of differentiation and organization in systems
of action. What happens is rather, we suggest, that the interpenetration
comes to utilize higher-order mechanisms of the physiological control of
behavior which are much less readily identifiable directly as mechanisms
of "overt" behavior. It seems likely that these are most centrally located
in the higher centers of the central nervous system and involve "field"
phenomena of dynamic equilibria rather than gross physiological move-
ments.

The phenomenon of interpenetration, which we wish to treat not
only in the present context but more generally, is closely related to that
of internalization. In a sense parallel to that we will employ later in
speaking of objects internalized in the personality, we can, I think, here
speak of psychological structures as coming to be internalized in the
organism. By this we mean that through processes of learning the struc-
ture of the organism must be modified in ways such that ordered re-
sponses to stimuli (both in perception and in response in the narrower
sense) can be produced without repetition of the learning experiences.

Psychological Theory in Terms of Theory of Action 651

With respect to perceptual content we speak of this altered structure as
the physical basis of memory, with respect to performance patterns, as
skill. It may be suspected also that the learned potentialities of pleasure
production (including., perhaps especially, erotic pleasure) constitute an-
other focus of this organic modification.

There is a sense in which this conception of the boundary inter-
change between psychological and organic systems and their interpene-
tration with each other constitutes an approach to the old mind-body
problem. We do not wish to stress the philosophical aspects of the ques-
tion. We do stress that once one has learned to avoid reifying analytical
systems and has understood that plural analytical systems are involved
in the same concrete phenomena, there need be nothing mystical about
what is meant by mind (i.e., a psychological system) as analytically dis-
tinguished from organism or body. When we add the conception of or-
ganization and its relation to processes of control, and the conception of
emergence in an evolutionary perspective, we have a scientifically in-
telligible frame of reference for analyzing mind, body, and their rela-
tions to each other.

The Object-relations of Psychological Systems

Let us turn now to the second main set of boundary processes of the
psychological system, namely, the interchanges with objects external to
the organism-personality system. On general theoretical grounds I shall
maintain that, as the psychological-organic boundary relation involves
primarily an interchange of facilities, so the psychological-object inter-
change involves primarily an exchange of rewards. This, of course, can
be strictly true only if the object in question is itself an action system.
Hence for present theoretical purposes, we must regard the physical ob-
ject as a special case, because the relationship established between it and
the psychological system is one-sided; what for the psychological system
is a reward is, for the physical object, simply a "state of affairs." In other
words, the goal-attainment boundary of a psychological system — so long
as it is independent, not a subsystem of a larger psychological system —
defines its relations to a social system in the technical sense of the theory
of action.

What, then, can we say about the relations between output and in-
put in this interchange? Both inputs and outputs at this boundary are,
we have suggested, maximized in what has been defined as the goal-
attainment state of the system, the establishment of an optimum rela-
tion to the significant object in the situation. To use a term which has
both social and psychological connotations, we may say that the system
gains "support." As long as an optimal relationship with the object is
maintained — whatever the respective shares of system and object in its

652 TALCOTT PARSONS

establishment or maintenance — from the point of view of the "needs"
of the psychological system, events in the environment are supportive of
processes in the system. Put in slightly different terms the gratification
level is at a maximum — for this system in this situation.

In terms of its meaning for the system, however, the input of support
must be evaluated. This in turn involves a possible breakdown of the in-
put into components — a breakdown which of course need not be ex-
plicitly performed by the actor. The principal components seem to be as
follows:

1. The input of immediate goal gratification independently of any
conditions on which its continuance, repetition, etc., may rest.

2. A icomponent of predictable (and possibly controllable) condi-
tions in the situation, independent of any features of the relational tie
between ego (the psychological system of reference) and alter (the ob-
ject) on which the continuance and/or repetition of the gratification
may depend.

3. An integrative tie between ego and alter by virtue of which they
may be treated as belonging together in the same solidary collectivity in
the sociological sense.

4. A shared system of cultural values which define legitimate expec-
tations in the relationship.

It is as a resultant of these four factors that the attainment of a goal
is not merely "enjoyed" but positively valued. The relative weights of
these components will vary for different psychological systems and on
different situations.

Psychology calls the principal category of relevant output, goal at-
tainment, achievement, etc. Perhaps the best term is achievement. Here
the primary emphasis is on the agency of the system, on its "decisions"
or "commitment" to the goal state as a factor in bringing about that
state. The balance may vary from a situation where ego merely "enjoys"
a goal state freely "presented" to him without effort or foresight on his
part to a goal state which he "succeeds" in attaining only in the face of
the most formidable obstacles, but this element is always present.
Achievement in this sense can also be broken down into the following
components: (1) simple "acceptance" of the optimal situation as grati-
fying; (2) manipulative control of the conditions of the optimal situa-
tion which are independent of the relational tie to alter and of the com-
mon values they share ; ( 3 ) maintenance of the integrative tie by virtue
of which ego and alter are bound together; (4) conformity with their
shared values.

In so far as the object relationship is one to social objects, the at-
tainment and/or maintenance of the gratificatory state is subject to a
double contingency [cf. 22, chap. 1]. The first two sets of inputs and

Psychological Theory in Terms of Theory of Action 653

outputs are relevant to relations to a nonsocial object; the last two are
always involved in a social relationship. The double contingency con-
sists in the fact that ego's attainment of a goal is contingent not only
on his own actions in relation to a nonaction situation, but also on the
reactions of alter to ego's actions and their consequences for ego's goal
attainment.

It is also important to recognize that, in the general terms we are
using, the seeking of pleasure, so far as it serves as a motive, is not the
same thing as goal gratification. Pleasure is a state of the organism in re-
lation to the psychological system; goal gratification is a state of the
personality in relation to the external object system. Of course once a
psychological system has become firmly established it can learn to treat
the arousal of pleasure sensations from its own organism as a goal, but
only by treating the body as an "external" object.

For reasons which should be evident, the main tradition of experi-
mental psychology has been concerned only with the first two com-
ponents of "support" and goal attainment, namely, the gratificatory
relation to the object (primary reinforcement) and the "conditional"
factors most closely related, such as timing, periodicity and aperiodicity,
and the like. Where, however, gratificatory behavior has been studied
directly in relation to social objects, the other two factors have emerged
into prominence, most conspicuously in the case of Freud and his in-
tellectual descendants. Here factor c, the relational tie, appears mainly
in the form of object cathexis; the sharing of common value is prom-
inently involved in Freud's concept of identification.32

Because of these considerations, one must infer that the more general
case (in a theoretical sense) of the relation of a psychological system to
external objects is that in which the most significant objects are social
rather than physical. Freud was doing more than following his own
special interest in human socialization when he put relations with social
objects in the forefront of his thinking. But not only is this true, it is
necessary to go one step further. To simplify the exposition, I have used
as an illustration the situation where the social object is an individual

32 Cf. "Social Structure and the Development of Personality" (Psychiatry, Nov.
1958). For reasons we will take up later, it can be inferred that either where
object-relations are virtually confined to nonsocial objects or where such relations
to social objects as exist are highly stereotyped by instinctive patterns, the
potentiality of development of psychological systems is severely limited. The main
path to the development of the human level of personality is the introduction of
elaborate processes of socialization, i.e., of learning through interaction with social
objects who are bearers of a highly differentiated culture. This is a principal
reason why the empirical generalizations derived from the study of animal behavior
in nonsocial situations apparently are less fruitful for the psychology of the human
personality than it has often been claimed they should be.

654 TALCOTT PARSONS

person, an alter to the ego of reference. Human beings, however, are
not generally oriented to and integrated in object-relations only with
discrete individual alters. Their relations are to systems of social objects.
Ego's integration in such a relational system is integration into, the ac-
quisition of membership in, a social system, in the type case a collec-
tivity. The objects significant to him include not only the other members
of the collectivity as individuals but the collectivity itself. Freud's con-
cept of identification, for instance, must be taken to include reference to
the collectivity. For example, the main identification which occurs in
the oedipal period is primarily an indentification with the family of
orientation as a collectivity, not merely with the father or mother as
individuals.33

These considerations bring us again to the meaning of interpenetra-
tion in reference to psychological and social systems and the crucial con-
cept of internalization. It seems clear that physical objects become in-
ternalized, that there is on both the psychological and physiological levels
some kind of enduring structure which corresponds to every category
of objects of which the individual has had experience — without such
a postulate the phenomena of memory and the continuities of be-
havior could not be accounted for. In so far as object-relations are
learned, these structures must be built up and changed through processes
at least partly psychological.34

The same general principles apply to the internalization of social
objects, with the modifications which result from the double contingency
of social relationships. Double contingency is a critical source of in-
stability which makes the psychological certainties attainable in relations
to physical objects difficult, if not impossible, to reach for social objects.
Yet at the same time, the human socialization process seems to depend
on the double contingency of social interaction.

Apparently, the main reason for this is an "artificial" stabilization of
the environment by social interaction. This creates for the child a spe-
cially stabilized environment between the physical world and the nascent
personality system. This artificial environment is more closely adapted

83 There is another sense in which S-R-S theory may constitute the more
general case, namely, that all psychological systems involve elementary units on
this level. Hence in some sense, all higher organizations can be said to be
"composed" of such units. But we treat organization as an independent variable
in our system, so it is out of the question to derive the properties of the higher-
level systems simply from the properties of elementary units. On the other hand
to say it is the units plus their relations begs the question since it is in the category
of relations that the factor of organization is found, and its value must be specified.
For an elementarist theory these must be treated as parametric.

34 On the general nature of internalized object systems with special reference
to physical objects [cf. 20, Chap. 5],

Psychological Theory in Terms of Theory of Action 655

to the child's psychological need structure than the physical world; hence
it sharply modifies the conditions of learning.

The conditions of human learning are certainly complex; no one
factor alone is sufficient to explain them.35 Nevertheless, we may list four
important conditions. First is a stable identification, or set of identifica-
tions, with one or more social objects, which entails a cathexis or attach-
ment to the social object and the expectation of support (in some contexts
"love") in return. A second condition is sufficiently severe frustration of
previous expectations to disorganize previously established behavioral
patterns. A third is the selective rewarding of trials in accordance with
the expectations of proper behavior for the "higher" level of socializa-
tion. A fourth is the application, relative to this level, of a systematically
organized pattern of sanctions over a long enough period to furnish re-
inforcement adequate to bring about internalization.

Through socialization processes of this sort the social object system
comes to be internalized, not as one such system once and for all, but as
series of progressively increasing complexity succeeding each other over
a long period. What is internalized is a complex system of "expecta-
tions." These include more than definition of and motivation to expected
behavior for ego alone. In view of double contingency, this could not be
sufficiently specific, since how ego is expected to behave is always con-
tingent on how alter has just behaved and vice versa. It cannot then be
just a pattern of expectations for ego which is internalized but, as G. H.
Mead clearly saw, a reciprocal role-relationship pattern, including the
general norms governing the behavior of both ego and alter, or of a still
more complex system. In other words, there must be organized in ego's
psychological system a structure which corresponds to a continuing sys-
tem of ego's learned complementary role relationships in social collectivi-
ties in interaction with a plurality of alters. That structure is most defi-
nitely organized on a variety of axes through generalization of patterning.

The main structure of the human personality may well be organized
about the internalized social object systems as residues of the socializa-
tion process.36 Probably this is not, in the same sense, true of animal
personalities. The difference is linked to the far higher degree of con-
trol of human behavior by generalized, and hence abstract, cultural pat-
terns.

Furthermore, not merely the cognitive side but also the motivational
side of the personality is arranged around internalized social objects.

BS One attempt to analyze them has been made by Parsons and Olds [see 25,
Chap. 4].

38 The authority of Freud, in his later phase, can be claimed for this proposition.
Cf. especially The Ego and the Id [8], where he speaks of the ego as consisting in
the precipitates of cclost objects."

656

TALGOTT PARSONS

Not instrumental skills alone, but the goals of the mature personality are
organized through identifications and the consequent intemalization.
For the goal structure of the human adult cannot be derived from the
structure of the "instinctive/5 i.e., genetically inborn, needs of the or-
ganism.37

The discovery of the intemalization of social objects must be re-
garded as one of the most crucial in modern psychology. On the psy-
chological side certainly the main credit must go to Freud. It seems,
however, that Piaget arrived at it independently, especially in his early
work in the distinction between "moral realism" and "cooperation" [cf.
30]. Moreover, it is significant that, in somewhat different versions, it
was also clearly set forth by G. H. Mead, a philosopher who has also
been thought of as a social psychologist, and by a sociologist, Durkheim.
From the standpoint of systematic analysis, intemalization constitutes the
principal link between psychological and social systems.

A particularly important consequence of the discovery of the con-
cepts of intemalization and identification is the insight that the role-
expectation pattern must be understood not merely as a component of
the structure of social systems, but also at the same time as an establish-
ment of the personality, a part of its structure; i.e., they interpenetrate.
This will be discussed further, after an outline of the internal structure
of psychological systems has been presented.

Psychological and Cultural Systems

Input-output relations, interpenetration, and intemalization can be
outlined and partially analyzed in relation to a third major boundary of
the psychological system, namely, the boundary with the cultural, as
distinguished from the social, system. Just as for many purposes of "be-
havior psychology" organism and psychological system have not been
distinguished, so for many purposes in the present area social and
"cultural (II)" systems have not been distinguished. For more refined
analytical use, for the analysis of more complex systems, however, this
distinction becomes essential.

On the theoretical grounds referred to in the two preceding sections
I shall state, but not attempt to justify, that the main significance of the
interchanges between psychological and cultural systems is mutual
integration. Thus the inputs and outputs are neither facilities nor re-
wards but rather, regulatory "cues" which have a primary bearing,

37 These themes are much more fully developed in [25] and in my paper
"Social Structure and the Development of Personality," op. cit.

Psychological Theory in Terms of Theory of Action 657

not directly on the functioning of the system in relation to its external
situation but on its internal integration. They facilitate (or obstruct)
the adjustment of the units within the systems to each other.

The primary role which culture plays in the psychological system
is that of legitimation. As a result of this stabilizing factor, the system's
functioning is made subject to normative patterns. Thus, the culture
defines the conditions of stable equilibrium in accordance with "ex-
pected" conditions. In other words, cultural values are parameters which
establish certain perceptual and action thresholds and other forms of
selectivity.

Legitimation can also be seen as the aspect of the organization of
psychological systems most closely associated with the concept of "ration-
ality.35 Rationality may be conceived as organization of a psychological
system in accord with a system of norms so that, in specific situations,
it can perform in accordance with those norms. The legitimation input
(into the psychological system) may be divided into subcategories of
norms: (1) adaptive-cognitive rationality, where the correctness and
generality of the adaptive orientation provide the normative criteria
(i.e., knowledge); (2) instrumental goal-directed rationality, where
effectiveness in the attaining of specific goals is the criterion (principally
skill); (3) integrative rationality, where the internal harmony of the
psychological system itself is the criterion, giving each of its units and
subsystems an "acceptable" place (thus minimizing "defensiveness" ) ;
(4) "moral" rationality, where conformity with a set of norms transcend-
ing reference to the psychological system in question is the criterion.
Ordinarily these moral norms will be the cultural values institutionalized
in the social system (s) of which ego is a member and which are shared
with other members.

Turning to the output side, we may say that the general output cate-
gory from the psychological to the cultural system is "motivational com-
mitment." This goes beyond "understanding" of the relevant norms to
"acceptance" of them as guides to action in particular situations. The
capacity for such commitment is attained only through full internaliza-
tion of the norms; i.e., the norm is not simply an aspect of a pattern of
sanctions imposed by external objects but becomes a part of an internal
regulatory mechanism of the personality system itself. Motivational
commitments can be classified on essentially the same basis as types or
components of legitimation.

Particularly in the earlier stages of socialization the internalization of
norms proceeds empirically as part of the same process as the internaliza-
tion of social objects; indeed since the social object system is culturally
organized and controlled, it is impossible to internalize one without the

658 TALGOTT PARSONS

other. In the later phases, however, It seems possible to differentiate the
two functions in social interaction, and there is evidence that they also
become differentiated in internal function in the personality.

Probably after what Piaget calls the stage of "moral realism," the
individual comes to be capable of discriminating a rule from an aspect
of a concrete solidary or antagonistic social relationship. He becomes
capable, as we say, of treating the individual "impersonally" as a "case,"
or in more technical terms, by universalistic standards. This we may
regard as the differentiation of what is primarily a cultural reference
from the matrix of a social system reference.

Freud's distinction between the ego and the superego is the best
point of reference for an attempt to distinguish internalized cultural
patterns from internalized social objects in the personality. Freud con-
ceives the ego as composed of the precipitates of lost social objects. It
is also that sector of the personality most directly governed by the "reality
principle." This is quite consistent, for Freud considered the social
object system as the crucial component of reality.

The superego, Freud says, centers on the "parental function." [Of.
8, 9.] It originates when the child's family of orientation is internalized
as a collectivity, but precisely in its governing aspect. We may then
surmise that once this familial object has been "lost" in Freud's sense,
it is the system, of norms symbolized and implemented by the parents
which becomes the focus of organization of the superego. It is note-
worthy that for Freud the superego does not begin to function fully
until the latency period, when the child is emancipated from his family.
This coincides broadly in timing with Piaget's formulation. It is our
view that this differentiation and its maintenance require a special mode
of social interaction mediated by highly generalized cultural symbols.

Finally, we may note one further aspect of the relation of a psy-
chological system to its situation or environment. This is the residual
category associated with the pattern-maintenance function, not as it
operates through the integrative channels we have just discussed, but
directly through congruence or incongruence of values. Any concrete be-
havioral sanction would, as we see it, be resolvable into the three com-
ponents already reviewed as facilities, rewards, and legitimation sanctions.
The significant sanction form here is purely attitudinal — as to whether the
system in question "fits" into the larger value complex of which it is a
part. There is thus an absence of expectations on either side, for per-
formance or for sanctions specifically contingent on one another. What
is involved is rather an over-all "judgment of worth" which we may call
an attitude of esteem (or disesteem). This is the level of regard or re-
spect in which the system (most clearly in the case of a total personality)
is held by other social objects. Change of status in this respect calls for

Psychological Theory in Terms of Theory of Action 659

more than specific types of acts; it calls for a change in the structure of
the personality system, including its value content.

The Internal Structure and Processes of Psychological Systems

This brief review of the principal boundary processes of a psy-
chological system vis-a-vis nonpsychological systems can serve as the start-
ing point for systematic treatment both of the internal structure of the
system and of the mechanisms involved in its functioning. We will take
up these problems on two levels, first that which abstracts from internal
differentiation and second that which attempts to deal with its explicitly.

In the outline of the general structure of the theoretical system we
are using here, it was pointed out that the value system served as the
general point of reference for analysis of structure and processes in the
system. This has been the case in the discussion of boundary processes
between a psychological system and the nonpsychological systems of its
environment with which it interacts. That discussion did not take account
of variations in the content of psychological value systems. This will be
the task of the present section.

A psychological value system is the main point of interpenetration
between culture and the psychological system, in the most interesting
case, personality; it is defined as internalized cultural pattern. On the
cultural level a psychological value system must be placed in a frame
of reference which defines the category or population of higher-order
systems within which the particular range of variation falls. The terms
we will use, referring as they do to relative primacy of function in the
system of reference, are always relative, not absolute. We may thus say
that, within the given reference system, personality or culture A is more
adaptively oriented than personality or culture B. But we should never
state or imply that there is, in the present theoretical system, an
absolute standard which defines "the" adaptive value system, or a
cognate standard for any other type.

Differences between values in psychological systems can be stated
as differences in the relative degree to which adequate or satisfactory
solutions of the four functional problems have been emphasized. This
is because, relative to a given system situation definition standards of
adequacy for each type of solution are built into our theoretical analysis
of the types of input-output balancing. Thus, given ego's paramount
goal, estimation of the degree to which this is attained in a given state of
the system is not arbitrary but in principle can be precisely defined
(whatever the empirical difficulties) .

A somewhat related statement is that the value system is the focus
of a set of cybernetic control mechanisms which regulate the relations

660 TALCOTT PARSONS

between the given parts, essentially by assigning priorities whenever a
conflict situation arises or becomes a serious possibility. Hence, the most
fundamental feature of a value system is the order of priority it gives
to the solution of each of the four system problems. In interpreting this
statement, however, it is necessary to be clear about two things. First,
there are always limits to the extent to which any of the four problems
can be subordinated or neglected; we have defined all of them as
functional exigencies. Thus the denial of all goal gratification would
be as realistically impossible in the long run as would total neglect of
adaptive imperatives; it is always relative importance which is in ques-
tion, never whether or not an interest shall be respected to some degree.
Second, the realistic possibilities for rank-ordering values are always a
function not only of the kind of system involved but also of the situations
in which it is placed. Since situations may be presumed to vary over a
considerable range, the value system never can fully prescribe what the
priority relation must be within any short time period unless it is to
develop a rigidity in the total system which is incompatible with capacity
to cope with variability in the situational conditions. During longer
periods there may be structural change in the system.

Rank-ordering of functional problems presumably entails giving
relative "weights" in the determination of action of the system to the
principal types of different performance disposition and of sanction need.
By definition the largest weight is assigned to the paramount functional
need-disposition, and so on down. Therefore a typology of psychological
systems38 resting on types of values must give first attention to the need-
disposition which occupies top place in the rank order; the range of
variability will become progressively smaller as a function of variations
in each of the others down the line until the fourth is reached. At the
same time, it is most important to note that the relative weight of the
lower-order functions is never completely determined by their position
in the rank order. We /can say only that this position sets certain limits
to the variation; it cannot exceed an upper limit without changing
places with the unit above, conversely it cannot exceed a lower limit
without exchanging places with that below. This consideration intro-
duces an important element of flexibility into our classification. This
flexibility does not, however, destroy its validity, since shift in rank
order presumes certain qualitative changes in the characteristics of the
system.

By adopting this approach, certain combinatorial possibilities in the
theoretical universe we are dealing with provide a framework of analysis.

38 This analysis in principle applies to all levels of psychological system. It is,
however, easiest to illustrate it at the personality level and we will hence couch
our discussions in those terms.

Psychological Theory in Terms of Theory of Action 661

If attention is paid to the first-order function then there are four
"primary" types of value system; each of the four system-problem
solutions holds the position of first priority in one of the four types.
Each of these types can then be subdivided according to which of the
remaining three holds second place in the priority scale., yielding twelve
possible "secondary types.35 Finally, each secondary type can be divided
into two according to the relative priority of the remaining two pos-
sibilities, yielding twenty-four "tertiary" types. Since we assume a uni-
verse where each type is characterized by the rank order of four com-
ponents, this exhausts the combinatorial possibilities, since last place is
fully determined when the other three are given. However, it should
be remembered that there is a further range of flexibility open in the area
of range of "weights" within a given place in the rank order.

We have held all along that a primary feature of a value system
as the focus of internalized culture is its property of generalization. In the
present context, we interpret this to mean that the type of solution
arrived at for any one of the four system problems will be generalized
to the other three; the orientations toward all four will tend to con-
stitute a meaningfully coherent system. We would hold that this system
provides the central structure of a personal or other psychological value
system. It is relatively easy to spell out these implications for first-order
values and hence the broad characterization of primary-value types. Be-
cause of the refinement and subtlety of the distinctions which have to
be made, spelling out implications becomes increasingly difficult as we
proceed to the levels of secondary and tertiary typology. In all this, the
problem of the range left open for "variation" must also be kept con-
tinually in mind.

Given that the locus of the primary value pattern is in the latency
or pattern-maintenance aspect or "part" of the system, our problem
is to spell out the implications of a given type of content of this "cell33
(in the formal paradigm) for the other three functional contexts and
input-output balances. Throughout this discussion it will be obvious that
we assume that the primary pattern tends to be maintained; the question
is how.

Let us start with a system which tends to maximize goal-attainment
values. Remember that we assume that the goals of the mature per-
sonality are for the most part learned goals. Our present analysis is not
concerned with the content of these goals; this depends on the kind
of socialization process the personality in question has been through. Our
present concern is with the implications of primacy of attainment or
gratification of whatever goals have been learned over other system
functions. In the goal-attainment function, this will give a primacy of
functional performance and/or gratification ; in this particular sense, we

662 TALGOTT PARSONS

may speak of a primacy of interest in gratification or in power.39 But
it should be kept clearly in mind that given the goal content, this is true
of any goal-oriented activity. What is peculiar to this value or structural
type is not this, but the primacy of goal-attainment values, hence of
interest in power including both the performance aspects and the
gratification aspects, over the values of other system functions.

Psychological discussion of cognate problems has tended to em-
phasize the interest in goal gratification (and often its organic counter-
part, pleasure). But if attention be shifted to the relational aspect and
the concern with alter as the source of gratifying sanctions, this primacy
leads, at least in one sense, to an interest in power, in capacity through
effective performance to control alter's behavior in the interest of ego's
goals. The balance between the gratification and the power emphases
then will be a function of relative activity or passivity in handling
situations, i.e., the degree to which ego attempts to control the situations
in which he seeks goal consummation as against the degree to which he
accepts them as they come.

Second, with respect to the adaptive function the goal-oriented
personality will tend to give primacy to "practical know-how55 for the
attainment of his given goals. He will then tend toward a strong
pragmatic-adaptive interest with a complementary limitation on the in-
terest in generalization of facilities. Ego will tend to feel that adaptiveness
beyond the limits of his own goal interests is not of "any use." He will
not maximize the adaptive function, but will tend to subordinate it to
goal attainment.

Since this psychological type is primarily concerned with external
situationally oriented goal attainment, his handling of the integrative
problem of his personality will tend to be subordinated to this interest.
The most important consideration here is that, however important the
conception of a paramount goal for the system as a whole may be, and it
certainly is high, concretely there will always be a plurality of par-
ticularized goal interests which compete for gratification and which
cannot all be satisfied at once or all of the time. The primary integrative
focus then will be the rationing of gratification opportunities among
competing goal interests. Many such interests will be inhibited at any
one time, and all of them at some times, in the interest of optimizing
the gratification balance of the personality as a whole. Put otherwise, this
involves hierarchization of goals, i.e., a rank-ordering of the rights of dif-
ferent subgoals to gratification. There is, it may be noted, an important

89 A special case of the potency referred to above. This presumes that
power is generalized capacity to mobilize the resources of a system for the attain-
ment of a goal. Personality and social system references of the concept should
not be confused.

Psychological Theory in Terms of Theory of Action 663

relation to the adaptive problem in that gratification opportunity is a
function of control of the relevant objects. Gratification can be jeop-
ardized either by the competition of another goal interest or by the
refusal of the object to "cooperate." Optimization requires some balance
in coping with two sources of danger which are largely independent of
each other.

Finally, what the pattern-maintenance function itself? The principal
"drive" of a goal-oriented system is to attain and retain the consum-
mately state. But because of the inevitable factors of instability in the
situation and perhaps within the system, for the sake of long-run
maximization or optimization of gratification, consummatory interests
must suffer renunciation or inhibition. Inhibitions or renunciations may
be called for in the face of adaptive and integrative exigencies; the
pattern-maintenance function regulates the relationship between the other
three and determines which shall take precedence on a given occasion.

But there is a further pattern-maintenance problem, that of main-
taining commitment to particular goals. Probably the most important
threat to pattern maintenance within the system lies in certain conse-
quences of personality development; namely, that both values and goal
structures have had to be built up by a process of socialization. Hence
the residues of earlier value- and goal-commitments are still to some
degree operative ; under strain there is always the possibility of regression,
i.e., of abandonment of more mature goals in favor of reactivation of
earlier ones. Control of tendencies toward regression is a primary pattern-
maintenance function in the personality.

We have suggested that the rank order of the other three value-
pattern components below the one enjoying primacy is intrinsically open
to the whole range of combinatorial variation. Yet as a function of the
intensity of commitment to one primary value type there may be a
"strain" in the direction of favoring one subsidiary rank order over
another. In the present case, the central consideration appears to be
the high significance of situational relations. This would tend to give
second place to the adaptive function, subject to the limitations we have
indicated. This tendency could be counteracted, for instance, in the case
where maintenance of commitment to the goal pattern in question in-
volved sufficiently serious internal strains, which would tend to give
second place either to the pattern-maintenance or the integrative
function.

As a second major value type, let us consider that based on the
primacy of adaptive values. This, it will be remembered, will involve
interest in control of the situation of action, and in the generalization
of the means of control relatively independently of particular goals. With
respect to goal type, primacy of adaptive values will tend to a kind of

664 TALGOTT PARSONS

goal we may call that of achievement or success. This means essentially
that subsidiary goals will be evaluated in terms of their contribution to
the larger goal of putting ego in a position to achieve any goal he may
come to be interested in, with the specification of particular goals left in
abeyance. We may speak of this position as that of generalized mastery
extending over the whole range of actual and possible situations. In the
interest of this generality of mastery, it is necessary to avoid too many
and too deep commitments to particular goals and particular solidarities.

On the positive side, as generalized means to success, two categories
seem to be of primary significance. Internally the generalized means of
control of situations is knowledge, particularized in the form of skills.
But knowledge is essentially the capacity to understand and prepare for
anything which may happen in the external world. Externally, the central
category is wealth which, in a highly differentiated society, is the most
generalized means of influencing others' behavior in the desired direction.

On the negative side, with reference to goals and otherwise, the per-
sonality with primacy of adaptive values will tend to be characterized by
high levels of discipline. He will be wary of expending his resources on
gratification of particular goal interests if this tends to impair the
maneuverability of his position with reference to alternatives. He will
also tend to be wary of commitments to solidarity since this also im-
pairs his freedom of action.

From this, it follows that ego's pattern of adaptive action will be
characterized by a relatively active disposition to exploit opportunities for
improving his adaptive position; it is in this direction that his primary
goals will lie. His integrative pattern, on the other hand, will be char-
acterized by a kind of utilitarian self-discipline, a strict control not in the
interest of harmonization as such but of adaptive efficiency.

The pattern-maintenance problem will have a similar but in some
respects different meaning in this case from that of goal-attainment
primacy. It is similar in that the maintenance of pattern is secondary
to other interests, and the focus of these other interests is on relations
to the situation external to the system. It is different with respect to the
significance of particular consummately goal states. The essential point
is that generalization in the situational reference tends to be maximized.
Once such a value system has become established there is less pressure
to attain particularized consummatory states, and the question of con-
sistency with an internalized pattern is more important. Perhaps we may
say that, for the goal-oriented type, the primary focus of integrative
strain is the problem of "expediency,33 namely, whether all-out consum-
matory commitment is justified in the particular instance. In the adaptive
case, on the other hand, it is the problem of the organization of the

Psychological Theory in Terms of Theory of Action 665

adaptive system as a system, since the essential focus is on the total
level of adaptive success, not on any one component contributory to it.

Can we say anything about the strains toward a preferred rank order
of the remaining three value components in the case where adaptive
interests are primary? The same basic consideration which was noted
in the goal-attainment case seems to apply here also, namely, the high
importance of the relation to and control of the external situation. This
would indicate the probability that other things equal, second place
would go to goal-attainment values. Probably, however, pattern main-
tenance would be a closer competitor than in the case of goal-attain-
ment primacy, because of the great importance of cognitive orientation
and of organized consistency in the adaptive system. Too strong con-
summatory interests, because of the particularistic relations to external
objects that they imply, would tend to threaten this consistency.

In both pattern maintenance and integrative primacy, the primary
orientation of the system is inward rather than outward to situational
objects. Hence, objects are more likely to appear as potential sources of
disturbance and less likely to be sources of opportunity than in the other
two cases.

In the pattern-maintenance case, the primary concern is with main-
tenance of the integrity of the internalized value system itself. According
to its content there will be more or less need for active "realization."
Where goals are specified, as they must be, the measure of effective per-
formance will not be so much the fact of attainment of a consum-
matory relation to the goal object as the meaning of this attainment in
terms of the value pattern. There is, therefore, a tendency to dichotomize
consummatory situations, in realization and in expectation, either as ex-
pressive realizations of the value pattern or as in conflict with it. The
latter can be justified only in terms of expediency and this is the primary
integrative problem for such a value system. The tendency, therefore, is
to accept certain goals, and pursue them with great energy, and to reject
others as totally unworthy. The person with high primacy of pattern-
maintenance values is likely to be an uncompromising "idealist" who
defines situational objects in black and white terms as either proper goal
objects or totally inadmissible.

When we come to the adaptive problem, the distinguishing feature
of the pattern-primacy personality is the minimization of the importance
of successful adaptation. The rigid pattern of selection which he imposes
on possible goals precludes a strong interest in maximizing the generalized
potential for goal attainment. He wants to know what specific goal you
are committed to. He will go all out in acquiring means for approved
goals, but is very suspicious of any command of facilities where the uses

666 TALGOTT PARSONS

to which they might be put are not carefully specified. If he is a religious
person he wholeheartedly approves a wealthy church, but is very sus-
picious of wealth as such and its secular uses. Again, therefore, he tends
to vacillate between commitment to adaptive values where the goal is
approved, and rejection of them where the goal is not. The sheer fact
that certain facilities are inherently generalized presents a major dilemma
for him. This applies both to knowledge and to wealth.

The integrative problem, finally, will tend to be met by a pattern of
hierarchization. But this is of a different character from that involved
in the goal-attainment primacy type; its primary function is not to ration
gratification opportunities, but to ensure the ascendancy of the primary
values and to ward off threats to this ascendancy. The internal discipline
then is likely to be more repressive, indeed, to take actual advantage of
repression, than in the other cases. The essential principle is to minimize
the necessary concessions to need-dispositions other than those directly
concerned with the primary values.

It seems probable that in this case integration is likely to come next
to pattern maintenance in order of importance, because the first line of
defense of the value system is internal. However, if the value system
calls for a strongly active orientation to external objects it can be that
goal attainment will take the second place.

Lastly, let us sketch briefly the implications of the primacy of integra-
tive values. The plurality of need-dispositions will be frankly acknowl-
edged, and the primary effort will be to weld them into a harmonious
system so that they can "live peaceably together." There will be a strong
tendency to take the underlying pluralism for granted as grounded in
"human nature." The hierarchical aspect of rank order will be recog-
nized, but as a "natural" thing rather than as something to be enforced
for other reasons.

In the goal-attainment context the primary goal type will be what we
have called "satisfaction"; it will be the attainment of diffuse solidary
relationships to persons and collectivities so that their patterns of integra-
tion can be reflected in and reinforce that of ego's own personality. In
the adaptive sphere there would tend to be the least motivation to
positively active adaptive improvement and the greatest tendency to
"come to terms" with situational objects in such a way as to protect the
internal integrative balance. The primary exception to this would be
the interest in particular types of skills, primarily in the field of human
relations, which would facilitate the attainment of satisfaction goals.

The pattern-maintenance function would also tend to be relatively
subordinated, and above all, too rigid idealism would be avoided. The
value system would have to provide internal sanction to the pluralism of
motivational interests and give positive value to their variety and to the

Psychological Theory in Terms of Theory of Action 667

task of harmonizing them. There would be a premium on self-control,
not in the interest of a single motive but in that of moderation and
balance.40

It seems probable that second place will go to the pattern-main-
tenance function, since only relatively "strong character33 can provide
protection of the internal harmony ideal against the inevitable situational
pressures with their wide variations. In particular situations, however, the

TABLE 1. PSYCHOLOGICAL ORIENTATION TYPES

Mode of
meeting
exigency

Type of value system

Goal-
attainment

Adaptation

Integration

Pattern-
maintenance

Goal-attainment

Commitment

Generalized

Pluralism of

Commitment

to implemen-

success or

acceptable

to meaningful

tation of

mastery

goals

goals and

paramount

avoidance of

goal

expediency

Adaptation

Maximization

Interest in

Tendency to

Restriction of

of power

knowledge

compromise

generality to

(practical

and /or

with external

bearing on

know-how)

wealth

object inter-

values

for para-

ests

mount goal

Integration

Rationing of

Discipline in

Internal har-

Hierarchiza-

gratification

the interest of

monization,

tion: repres-

opportunities

active mastery

avoidance of

sion of con-

conflict

flicting needs

Pattern-

Control of ex-

Maintenance

Control in in-

Maintenance

maintenance

pediency and

of organized

terest of mod-

of integrity of

regression

self-control

eration and

the value

balance

position

Avoidance of

rigid idealism

appropriate type of goal attainment can become extremely important
where external reinforcement of the internal pattern becomes crucial or
is threatened.

Table 1 gives a schematic view of the relations between the first-order
value-pattern component and each of the other three just reviewed.
Each column represents a schematic description of one of the four
primary psychological value types; since there are four principal ways
of meeting each of the four system problems, sixteen combinations are
possible. The diagonal from left top to right bottom shows the points

40 The Confucian Chinese ideal of the "superior man" may be regarded as one
of the classical expressions of the value pattern for the case of personality.

668 TALCOTT PARSONS

in each case where the strongest motivation to fulfill a value requirement
will come, because this is the functional problem which is given first
place in the rank order for the type in question.

No attempt is made in the table to represent the more refined bases
of variation. It is, however, important to our general argument that
each of these primary types can, by the logic of rank-order analysis alone,
be divided into three subtypes according to which of the three remaining
value components takes the second place, and the resulting twelve can
again each be divided into two according to which of the remaining two
comes next. Furthermore, it should be remembered that, still further,
concrete variability is possible because a rank-order position of a given
component is interpreted to involve a range rather than a point. Since
we restrict the components discriminated to four, for most empirical pur-
poses it would seem that these ranges should be relatively wide. Finally,
it will be remembered that we emphasized the relativity of the para-
metric points of reference which would locate the total population of
personal value systems our scheme can deal with relatively to others.

These points we emphasize to forestall criticism that we are imposing
a classificatory strait jacket on what many feel to be the infinite variety
of psychological types, a classification which is necessarily unrealistic
because it cannot take account of the finer nuances of difference. We
have distinguished 24 possible types (by no means all of which are of
equal empirical probability), and each of these in turn may vary as a
function of differences in the population parameters and in position
of the variables within the range consistent with maintaining the rank
order. In putting forward this classification, we have still not taken into
account a further very important source of variability; namely, that
concerned with pathological factors.

If psychological theory is to deal with the problem of variability
in any theoretically determinate way, it cannot rest content with simply
asserting the fact of infinite variety. It must impose some order in terms
of a manageably small number of categories. We feel that the scheme
just outlined offers quite sufficient possibilities of complexity to occupy
theorists for a long time. Whatever other faults it may turn out to have,
gross oversimplification of the problems is not likely to be one of them
except in the sense that all scientific analysis oversimplifies. What we
have attempted to do is to derive a high order of complexity from vary-
ing combinations of a small number of elementary components.

The same pattern of analysis which we have used to develop a classi-
fication for normal psychological system types can also be used to
approach the problem of characterization and classification of the mecha-
nisms of personality functioning. For our purposes the first essential con-
sideration is the distinction between those mechanisms which mediate

Psychological Theory in Terms of Theory of Action 669

the relations of the total personality system to the situation external to it
and those which mediate internal processes between units of the system.
The former we will call the mechanisms of adjustment, the latter the
mechanisms of defense. It is clear that so far we have laid the ground-
work only for approaching the mechanisms of adjustment since we have
not yet dealt with the internal differentiation of the personality system.

The scheme we have been using provides a very simple basis for a
classification of patterns of adjustment. The essential point is that the
equilibrium of the system depends on the balance of inputs and out-
puts both in each category and in all of them taken together.

In the most direct sense, each of the balances can be affected by
ego's either increasing or decreasing his own rate of output. Thus in the
consummatory phase, ego may increase the intensity of his effort to
maintain the consummatory relation, presumably by performances
thought to be gratifying to alter, or he may decrease the rate of consum-
matory effort and thereby also decrease the input from alter by lessening
alter's "incentive" to continue the input at the same rate (which from
alter's point of view is output). Similarly, in the adaptive context, ego
may increase the rate of adaptive output by adding to the effort or
energy deployed in this direction, and by withdrawing energy or effort,
he may decrease the relevant output and, presumably through the effect
on alter, the corresponding input. Finally, the same would apply at the
integrative boundary in terms of increasing or withdrawing motivational
commitments.

Thus with respect to each of the boundary relations of the system we
have, basically, two possible directions of change, a positive active, in
one sense, "aggressive" direction and a negative passive "withdrawing"
direction. In any given case, the effect on ego's personality of such a
change of course depends on the shape of the function, namely, the
relation of the rate of addition or subtraction of output to valued input
rate. But also it depends on the shape of the function in terms of which
alter reacts to ego's changes of behavior. We are not in a position to
specify these functions on either side except to say that ego's curve of
"demand" for any category of input from alter will tend to slope down-
ward to the right, and the curve of his supply of any category of per-
formance in response to alter's actions will tend to slope upward to the
right.41

41 The assumption is that output is a quantitative function of rate of input
received for each level of output; equilibrium is defined as the point where certain
quantities coincide. The theoretical model which is most highly developed and
most appropriate to the present problem is that of supply and demand functions
as used in economic theory. The statement just made about the slopes of input and
output functions may be regarded as a statement of the law of effect. [Cf. 27,
Chap, ],]

670 TALGOTT PARSONS

These possibilities of increase and decrease of output respectively
apply to the three "open53 boundaries of a psychological system. These
may be said to constitute the elementary mechanisms of adjustment. But
there are two orders of possibility of organization of these elementary7
components into higher-order mechanisms. These have essentially to do
with the higher-order and the lower-order references respectively in the
hierarchy of normative control and of cultural generalization.

The higher-order reference concerns the importance of the sanction
of esteem, as we have called it. For ego, the meaning of a given alter-
ation in any one type of performance output may be not to secure a
specific change in the rate of the corresponding category of input from
one or more specific alters, directly, but to enhance the level of esteem
in which ego as total personality (or other relevant system) is held by
the significant alters (and also by mechanisms involving intenial
processes, perhaps his level of self-esteem). This enhancement of esteem
will depend on the same types of acts involved in the more specific
performance-sanction interchanges, but directed to a different and
more generalized (with respect to ego as a personality) meaning-level.
The meaning of esteem must be defined in terms of a value system
common to ego and alter, which is internalized in both their personalities
and institutionalized in the social system in which their interaction
takes place.

The lower-order reference, on the other hand, concerns the im-
portance and the possibility of control of the situation, i.e., of alter's
behavior. There is a sense in which what we have just called the elemen-
tary mechanisms of adjustment are to be regarded as instruments of
control because, in each input-output context in social interaction, alter's
behavior is always in some degree, contingent on ego's, as well as the
converse. But by control in a broader sense, we mean combinations of
the three output types (and also possibly of the manipulation of esteem)
in such a way that a desired pattern of alter's behavior involving several
of the input categories to ego's psychological system can be brought
about or made more probable. The most familiar case is adaptive
activity which is oriented to increasing the probability of consummatory
situations occurring, through "inducing'3 alter to enter into the consum-
matory relationship, or "coercing55 him into it.42

Control thus, as we conceive it, is the process of enhancing the input

42 By inducement, we mean the offering of rewards for alter's "cooperation,"
i.e., using control of the situation to make it more favorable to alter's interests
than it otherwise would be, contingent on his doing what ego wants. By coercion,
we mean the obverse case of threatening to use control of the situation to alter it
to alter's disadvantage if he does not do what ego wants. "Seduction," as the term
is used by psychoanalysis, is a special case of inducement.

Psychological Theory in Terms of Theory of Action 671

from alter's action (or reducing it if this is "desired") by indirect
measures, i.e., by operating on other elements of alter's input-output
balance than the one directly of interest to ego with a view to the effect
on one particular performance-sanction interchange.

Before turning to the internal differentiation of the personality
system, a word about the relation of the mechanisms we have outlined
to processes in the psychological system over time. We have suggested
earlier that if the situation of a system were perfectly stable and if its
internal state were stable (including the flow of energy from the
organism) the rates of input-output and their ratios would also be stable.
But this patently is never the case for long. The situation and certain
internal factors are continually changing and there must be "responses"
to these changes. But neither changes nor responses are random. They
tend to become structured in the form of differentiation of types of re-
sponse (in terms of relative primacy) at different periods in time, e.g.,
in some kind of cyclical periodicity. The diurnal cycle of wakeful activity
and sleep is probably one of the most fundamental of these; sleep cer-
tainly is partly a physiological phenomenon, but it seems to us most un-
likely that psychological factors are of negligible importance. In other
words, the differentiation of behavior into temporal phases which has
been clearly identified in small groups and in large-scale social phenomena
such as the business cycle, certainly applies to psychological systems as
well.

Determinate temporal phases result from the combination of two
fundamental considerations. First, the external factors affecting a psy-
chological system are continually changing. If to some extent the system
did not change its state in response to them, the effect of the range of
change in its inputs would be intolerably great and would be incompatible
with the maintenance of a stable organization as a system. Second, a
stable organization cannot change its state in nearly random fashion
without dissolution. If the effect of situational fluctuations on the system
were not partially neutralized by "active" mechanisms which partly
counteract these effects, it equally could not persist as an organized
system. The emergence of a pattern of changes of state over time may
be said to be the consequence of the fact that neither of these two
"radical" solutions is realistically possible; the actual process is a
"compromise." There must be some "give" in response to the in-
stability of the situation, but also there must be some resistance. Dif-
ferent types of psychological systems will presumably have patterns char-
acterized by different "amplitudes" of situation-determined fluctuation
and duration of phases. Thus, a "psychopathic personality" tends to be
too responsive to situational changes, a "compulsive" personality too
unresponsive.

672 TALGOTT PARSONS

The Internal Differentiation of Psychological Systems

Included in our definition of a system for purposes of the theory of
action was the proviso that it must consist in at least two interacting
units; otherwise it cannot be treated as a system but only as a unit in a
larger system. We also stated that systems of action tend to differentiate,
and though by no means exclusively, in the first instance, this differentia-
tion tends to take place on functional lines. Such functional differentia-
tion tends to define the goal orientations of the differentiated units or
subsystems, the rule being that the goal of the unit is a "contribution"
to the functioning of the system, tending to be specialized in one
principal function in the primacy of this function over others.

This functional basis of specialization with the system as a reference
point cannot alone determine the structural patterning of any system
without modification. For the units themselves are systems subject each
to its own functional exigencies in categories other than goal attainment
and the system as a whole is subject to situational exigencies which may
modify the "ideal" structure. Above all, the same order of input-out-
put relations and of interpenetrations obtains between units in their
capacities as subsystems as obtains between the larger system and the
situation in which it is placed.

To deal with all the ramified possibilities of modifications of psy-
chological systems at a functional basis of differentiation is beyond our
limits of space or competence. Furthermore, it must be noted that many
psychological systems do not attain a degree of differentiation where
even the four basic functions can be clearly discerned as foci of dif-
ferentiation. The best we can do here is to take a hypothetical four-unit
psychological system where the primary basis of differentiation has been
the four system problems as outlined in our general discussion of systems
of action. It is hoped that in this analysis of the psychological output and
input types involved we can identify some familiar mechanisms of psy-
chological functioning.

What we have called the functional basis of differentiation serves to
"locate" the unit in question relative to others in our action space. Be-
sides the coordinates of location, however, we specified that a unit has
a property of "potency,33 of relative weight in the system. We suggested,
too, that in the case of psychological systems what we mean by potency
can be identified as what Olds has called "motive force." Hence in
addition to the location of the units, an essential parameter of a system
is always the distribution of motive force between its units. This will
in turn consist in two components: what Olds's calls "intrinsic" mo-
tive force, which is relatively stable and can be redistributed only
gradually, and his "added" motive force [see 20, chaps. 3, 4], which

Psychological Theory in Terms of Theory of Action 673

Is more variable and continually redistributed in the course of system
processes.

It may be assumed that, in any system capable of Integrated be-
havior, the distribution of intrinsic motive force is unequal and that the
highest level of intrinsic motive force probably will be found in the unit
specialized in the function of goal attainment. This is specifically true
vis-a-vis the adaptive unit, as Olds has convincingly shown, but I think
it Is more generally true; it is a condition of capacity to "act." Indeed a
change in relative position, as in revolutions in social systems, simply
makes what was previously a "means" into an "end" and endows it
with the qualities of a goal-specialized unit.43

We may suggest that the distribution of the intrinsic component of
motive force, which is a stable property of the unit in question, belongs
to its "pattern" features (Olds refers to the "pattern force" of the
system) and in general is not involved in the input-output interchanges
between the units of the system. Hence, we can ignore it for the rest of
the present analysis (i.e., we treat its distribution as given and do not
attempt to analyze changes in it) .

It should be clear that the boundary interchanges between the units
of a system — i.e., the internal processes of the system — should be subject
to analysis in terms cognate with those employed in analyzing the inter-
changes between the system itself and the situation external to it. How-
ever, there are two differences. First, we are dealing with a different
level of system, the personality as a whole composed of intrapsychic
units — hence, the parametric references must be different from those
appropriate to the external boundaries vis-a-vis nonpsychological systems.
Second, we are dealing with all of the internal boundaries of a system,
not only the three external boundaries considered before. There should
be six interchanges between each pair formed by the possible combina-
tions of the four functionally differentiated units.44

First, let us take up the interchange between the primarily adaptive
subsystem and the primarily instrumental goal-oriented subsystem. This
is the relation in which, to use Olds's terms, "stimulation" is transmitted
"forward" from the stimulus-processing unit to the goal unit, and
"motive force" is transmitted "backward" from the goal unit to the
adaptive unit.

Stimulation (sometimes also called "expectancy"), on the intra-

43 In general Olds's concept of "added" motive force in the psychological sense
just defined is cognate with power (in the political sense) for macroscopic social
systems. It is interesting that the properties both of motive force and of power arc
perhaps in about the same stage of imperfect theoretical clarification. At any rate
the present discussion of motive force must be considered extremely tentative.

44 In this we follow the paradigm set forth earlier. [Cf. Fig. 2.]

674 TALGOTT PARSONS

psychic level, is cognate with organic energy on the level of input into
the psychological system. Generally in the psychological phase cycle, it
is a set of conditioned stimuli which are the external activators of this
system* What it does in relation to the goal-attainment system is to "put
at its disposal" control of the relevant cognitive mapping or planning
facilities which have been built up in the system.

The reverse flow is that of added motive force. By this in psy-
chological terms is meant the motivational capacity to perform the
additional information-processing job which is needed for the implica-
tions of this particular stimulus to be evaluated in terms of the motiva-
tional interests of the system. We may say that an expectancy for
gratification — indicated by the internal reaction to a conditioned stim-
ulus— stimulates internally the mobilization of the resources for the job
of planning how to get to the goal.

This interchange is a crucial one because of its relevance to a historic
psychological problem, namely, what mechanisms can account for the
"teleological" aspects of behavior? Some schools, of course, have held
that any "backflow" process was impossible since only antecedents in-
fluenced later events.

We feel that this view neglects the clear implications of a conception
that processes occur in a system of interdependent units; it fails to see
the possibility of treating the psychological system as a system of plural
units with mutual feedback relations to each other. It also confuses ante-
cedence of particular events in time with function in the system.

Once we see information processing, on the one hand, and commit-
ment to instrumental action, on the other, as the functions of two differ-
ent units in the system, there is no difficulty in conceiving of a shifting
balance of internal interchanges between these units over time in which
the output of each is instrumental in facilitating the other. It is essentially
a feedback system. The "backflow35 of motive force may be treated as a
feedback consequence of the activation of the adaptive unit by stimula-
tion and its communication of that stimulus "forward" to the goal unit.
Vice versa, the activation of the goal unit, e.g., as part of an internally
determined "need'3 cycle, may produce an output of motive force which
leads to "seeking" the appropriate external stimuli, and when they have
been found, passing expectations "forward" to the goal unit.45

The interchange of stimulation for motive force between adaptive
arid goal units is not sufficient to account for the determination of actual
courses of action of the system as a whole. A second vital set of processes

45 On the level of the macroscopic social system this interchange is cognate
with the investment of capital in the economy and the corresponding output of
"control of productivity." Capital, we hold, is a form of political power and as
such cognate with motive force. [Gf. 27, Chap. 2.]

Psychological Theory in Terms of Theory of Action 675

implies an interchange between the goaf unit and the integrative unit.
Here the problem is as follows : the goal unit is activated, we shall say,
by an expectancy of a particular gratification opportunity in relation to
an external object. But this is only one of many possible gratification
opportunities; there must be an ordering of the acceptance of such op-
portunities in time and as between the different "needs" of the system.
Essentially then, there is needed an adjustment between the unit
which has responsibility for commitments and the integrative unit which
must give "consent" or "support" for the commitment in terms of its
bearing on the internal balance of the system. In this interchange, we
suggest, the output of the goal unit is commitment or "decision to act,"
whereas the corresponding integrative output is "support." Apparently,
this means conflict arising from the commitment to accept or to re-
nounce a particular gratification opportunity is minimized through read-
justment of internal rewards within the system. We have reviewed several
types of such integrative process such as the rationing of gratification
opportunities. Rigid scheduling is another possible device.

Since relative to the organism, a psychological system is essentially
a set of mechanisms of control, we suggest that the main control of be-
havioral responses in the organically manipulative sense rests in the pat-
tern-maintenance unit of the psychological system. Using a sociological
analogy this is a "technical" function, and technical functions are located
in this subsystem. Hence what Olds calls the "response-control" unit may
be located here. The same basic psychological structure can, of course,
be the controller of the use of facilities outside the individual's own
organism, provided they are controllable. The basic psychological func-
tion is not operation of facilities but implementation of control; it is
specification and authorization of use, not the technical processes by
which a goal state is brought about.46

The psychological question then concerns the factors involved in the
determination of the utilization of the lower-order facilities available to

4(5 To take a particularly dramatic example, I happen to be writing this section
aboard a ship traveling from the United States to Europe. One of my goals
recently has been to "get to Europe." But since boarding the ship in New York,
and until I disembark on the other side of the ocean, I will not have "done" a
single thing which contributed to the physical transportation of my body across
the Atlantic. That is all done by the ship controlled by its officers and crew.
Indeed, I am strictly forbidden even to try to influence the process. Psy-
chologically, what I have "done" is not the technical action of transportation, but
to decide to go, to make a reservation, to pay for it, and to appear at the proper
pier in New York at the proper time. Then, once aboard, all I can do toward
this goal is to wait until the ship gets me there. I have controlled the relation
between the ship's operation and my own change of physical location, but in no
technological sense have I "transported myself" across the Atlantic.

676 TALCOTT PARSONS

(i.e., controlled by) the psychological system, notably, but not exclu-
sively, the responses of the organism. The first condition of the activation
of response facilities is adequate information about the situation. This
is given in the interchange between the pattern-maintenance unit and the
adaptive unit. Organized, processed information is received — and its out-
put by the adaptive unit is a function of the motive force which its ex-
pectancy-output has elicited from the goal unit. This information tells
(to some approximation) what response or other technical procedure
would be most effective in attaining the goal state.

What does the pattern-maintenance unit "produce" to balance
against its input of processed information? Essentially it puts the facili-
ties of the organism at the disposal of the adaptive unit for achieving
its functions; i.e., it controls the perception of external events and the
processing of information (this includes relating it to the stored and
catalogued information we call memory). I would thus locate memory
in the pattern-maintenance subsystem and in the pattern-maintenance
"cells" of the other subsystems.47

A second fundamental condition of the commitment of response
facilities to action is what may be called "integrative authorization."
Any commitment raises an integrative problem within the system since
there are competing claims for use of the inherently limited facilities
controlled by the pattern-maintenance unit. Adjudication of these claims
is essentially an integrative function; a social system analogue is the
legal system. On the personality level, I think it reasonable to suggest
that reference of proposed actions to the ego ideal is the kind of process
involved.

The reverse flow, from pattern-maintenance to integrative units, is
easier to characterize; it may be called the conditional commitment to
implementation. Essentially, this means that integrative authorization
has no practical significance unless there is a sufficient probability that
the authorized actions will in fact be carried out on the response level.
In other words, the control authorized must be genuine, not spurious.
Various psychosomatic phenomena, most obviously hysterical paralysis,
illustrate the failure of such control.

On grounds of general theory we may state, but again will not be
able to demonstrate here [cf. 24, chap. 5], that the pattern-maintenance
system has a dual set of functions. (Cf. under Psychological systems,
earlier. ) The first set, which includes selection of response and integrative
authorization, falls broadly under the heading of "tension-manage-

4TOn the social system level, this interchange is cognate with the input of
consumers' goods from the economy to consuming units — preeminently the house-
hold— and the corresponding input of labor services to the economy. [Cf. 27,
Chap. 2.]

Psychological Theory in Terms of Theory of Action 677

merit." These functions control tendencies to "act out" directly through
the command of response facilities within and outside the organism. But
the other set, which is relative to the pattern-maintenance function in the
narrower sense, has a special relation to maintaining the internalized
value system, and hence to legitimating the functions carried out by the
other subsystems of the personality.

This second function is paramount in the integrative interchange
between the pattern-maintenance unit and the goal unit. The essential
function of the pattern-maintenance unit here is legitimating the selec-
tion of a particular goal at a particular time by referring it to the internal-
ized values of the psychological system. It gives the go-ahead or author-
ization signal to the goal-commitment unit. But legitimation is not given
"for nothing53 ; the essential condition, i.e., the input from the goal unit,
is what we may call motive-force commitment, i.e., a commitment for
the transmission of sufficient motive force to the other units of the sys-
tem so that they can be motivated to perform their functions in the
implementation of the goal The goal unit is, we suggest, the main locus
of the input of motivation from outside the system in the form of goal
gratification. But it is not merely an input channel, but also an agency
of distribution; in fact, it does distribute from its "reservoir" but recoups
the loss when the goal state has been reached [cf. 20, chap. 4]. The com-
mitment of motive force by the pattern-maintenance unit is a "token";
it is not the main flow which goes to the integrative and the adaptive
units. But it is "assurance" that there will be enough motive force dis-
tributed to produce sufficient "effort53 to attain the goal.

There is, finally, one further integrative interchange, that between
the adaptive and the integrative units of the system.48 It gives the impetus
of adaptive, i.e., primarily cognitive, innovation as an input into the
adaptive unit and in exchange produces new "ideas." This interchange
may be the focus of "creativity" so far as it is internal to the psycho-
logical system.

To conclude this section, there should be brief comment on the
dynamics of process through time. Only in very short time perspectives or
at very high levels of abstraction do many important psychological proc-
esses appear to be phase movements, namely, stages through which a
process goes in the course of completion of a sector of behavioral process.

We have emphasized above our interpretation of goal-striving as the
result of a disturbance in the relation between a system of action and its
situation. With respect to any particular goal state there may be a typical
phase cycle beginning with state of latency (when, for internal or ex-
ternal reasons or both, the motive in question is quiescent), moving to-

48 This is cognate with the interchange of "entrepreneurial service" as an input
into the economy with new combinations of goods and services as its output.

678 TALCOTT PARSONS

ward an adaptive-preparatory phase, then from an instrumental phase
(of active striving) to a consummately phase, then to an integrative one,
and finally back to latency. The determination of the number of phases
and placing of dividing points may be arbitrary, but the main outline
is clear.49

These phase movements for whole systems can in turn be broken
down into interlocking phases for subsystems. Since Olds's analysis is the
most illuminating I know for psychological systems, I summarize some
main points. A unit, or as he says on the S-R-S level, a "concept," has
four phases: "latent," "conceptual" (adaptive), "perceptual" (neuro-
logical, "firing"), (instrumental goal-attaining), and "refractory" (inte-
grative) . The first step in bringing a motivational system out of latency is
possibly through external stimulation (CS) of the adaptive unit of the
system. The adaptive unit transmits stimulation to its successor, the goal
unit, as the process continues and the adaptive unit comes into the per-
ceptual phase; this is sufficient to conceptualize the goal unit. During this
phase which, for the system, is the adaptive phase there is active inter-
change between these two units. As the system passes over into the instru-
mental goal-attainment phase, the adaptive unit tends to become re-
fractory and the goal unit to pass over into the perceptual stage with
actual presentation of the goal object (US). After consummation, the
goal unit in turn becomes refractory and the system as a whole goes into
an integrative phase, etc.50

The essential point is that a system phase cycle is in one respect the
resultant of several different but synchronized phase patterns of its units.
In general, a unit reaches its consummatory goal-attainment phase
when its special functional problem has highest strategic importance for

49 This was first worked out for small groups by Bales and Strodbeck. It is
fully reported in [24, especially Chaps. 4 and 5]. Olds has applied it to the S-R-S
psychological system [20, Chap. 4].

50 Two types of factor may account for what Olds calls refractoriness, i.e.,
spontaneous cessation of consummatory activity on the part of the system or a unit
of it. Some kind of intrinsic "satiation53 process may set a limit. Stomach
distention as a consequence of eating would be an example. The negative stimula-
tion from this state will surely eventually outweigh the enjoyment from continued
eating. The second factor, however, is the demand of other motives in the psy-
chological system for gratification opportunities. We presume that giving the first
motive its opportunity has necessitated holding back others which displayed some
"tension." The longer this inhibition lasts presumably the greater the tension,
and this forces cessation of one type of gratification to make way for others,
including the instrumental activities necessary to attain them. The importance on
theoretical grounds of the principle of inertia should predispose us in favor of
the second line of explanation. It is always easy to relieve theoretical embarrass-
ments by making ad hoc assumptions about intrinsic satiation factors — but these
assumptions are generally to be regarded with considerable suspicion.

Psychological Theory in Terms of Theory of Action 679

the system as a whole, e.g., the adaptive unit is in the "perceptual" stage
during the adaptive phase of the system, preparatory to commitments.

The preceding review of the internal boundary interchanges between
functionally differentiated units of a psychological system is even more
tentative than other parts of this essay. It represents a first attempt to
adapt a paradigm which has been worked out in the analysis of social
system materials to psychological systems. As such it is certainly exceed-
ingly -crude. Yet it does seem to strike at a number of the critical prob-
lems in psychological theory and to illuminate them in certain respects.
It seems to me at least to "make psychological sense." If this first attempt
accomplishes only this, experience suggests that sufficient further careful
work will produce far better results. Unfortunately this takes time and no
results of such work can be presented at present.51

A Summary of Psychological Problems

As we have noted, our analysis of the structure and functioning of
psychological systems is necessarily crude at this stage. Careful attention
to our paradigm for action systems will show that the analysis made
here is systematic. But since it is couched in terms unfamiliar to most
psychologists (particularly in regard to the pattern of analysis in which
those terms are used), it may be useful to offer a brief summary of its
relevance to a few of the most familiar problems of psychology.

First, we may note that we make a clear distinction between a psy-
chological system and the systems external to it. We provide a frame-
work in which such concepts as stimulus and response can be interpreted
in terms of processes at the boundary of the psychological system : stimuli
are sources of input into the psychological system, responses, conse-
quences of a process of output from it. There is, further, a basis for the
discrimination of "conditioned" stimuli, or cues, and "unconditioned"
stimuli, or the presentation of goal objects.52 Similarly there is a basis
for the distinction between instrumental responses and goal responses.

C1 It might be appropriate here to adapt trie famous remark of Dr. Johnson
comparing the woman preacher and the dog who walked on his hind legs: the
remarkable thing was not how well the thing was done but the fact that it was
done at all. That a paradigm worked out in connection with sociological problems
could apply at all and make any sort of sense in the psychological field is the
remarkable thing. If it can be done, in time it probably can be done much
better. In my own experience, in another connection, at first it seemed remarkable
that a general sociological analysis could apply to the boundaries and internal
processes of the economy at all, but a good deal of work has produced a relatively
satisfactory fit in considerable detail, as documented in [27].

152 This must, of course, be treated as a relative distinction if we mean by an
unconditioned stimulus the presentation of a goal object, since it is to us a cardinal
tenet that most action goals have been learned.

680 TALCOTT PARSONS

Our treatment of the boundaries of the psychological system presents
three features which may be called somewhat unconventional from the
standpoint of most current psychological theory. First is the analytical
distinction between psychological system and organism. From this point
of view, stimuli (as inputs into the psychological system) are not events
in the environment of the organism, but the neural and other processes
in the organism which provide information to the psychological system
(it seems improbable that any stimulus influence can reach the psycho-
logical system except through organic sensory mechanisms). Similarly,
what psychologists ordinarily classify as responses are not strictly speak-
ing psychological processes but physiological consequences of psycho-
logical processes, controlled by the latter. We like Olds's reference to a
"response-control unit53 as part of the psychological system — it is control
which is distinctively psychological. On the organic side, the location of
the closest interpenetration of psychological and organic systems should
be not the skeletal-muscular system but the brain.

The second unconventional feature is the emphasis on the importance
of the social object system in goal orientation. We recognize that truly
psychological processes operate where the essential objects are purely
physical, e.g., food objects. But we hold that for our general theoretical
purposes this is a special case ; the higher levels of organization of moti-
vational systems involve the cathexis of social objects, not only as "indi-
viduals" but in social systems. Without the conception of identification
and of the internalization of social object systems, we do not see how it
is possible to account for the higher-level human organization of per-
sonality. We think we are authentically following Freud in this respect.

The third unconventional feature, if it can be called such, is the
analytical isolation of the relations of the psychological to the cultural
system and its discrimination from the social object system. We felt, for
instance, that this could help account for the importance of the distinc-
tion between ego and superego in Freud's later theory. We realize, of
course, that these last two problems, of social object system and culture,
do not arise at the simpler levels of S-R-S theory. But again we main-
tain that, seen in a larger perspective of psychological theory, the latter
deals with a group of special cases. Our scheme is quite capable of deal-
ing with the S-R-S case by suppressing certain ranges of variation with
respect to the organization of object systems and correspondingly with
reference to levels of generality of meaning.

Turning now to the internal aspects of psychological systems, by the
combination of a postulated (though avowedly schematic) differentia-
tion of the units of a system, and analysis of the input-output relations
between subsystems, we think we have found a place for dealing with a
number of other traditional psychological problems. First, we have dis-
criminated between the processes of consummatory goal-response com-

Psychological Theory in Terms of Theory of Action 681

mitment and those governing adaptive and instrumental behavior pre-
paratory to goal gratification. Here we follow Olds in the conception
that the goal unit of a system has a higher level of motive force, whereas
the adaptive unit has a lower level. It is in certain ways justifiable to
refer to the former as a "motive" and the latter as an "idea," but we
agree that this is a question of relative primacy; motives also have cog-
nitive elements, and ideas are also motivated.

Above all we think we can account for the apparent teleological or
"purposive" property of behavior in terms of the feedback relations be-
tween stimulation (expectancy) and motive force as outputs of these two
units of the system to each other. The phases of an action process may be
treated as coordinated but differentiated phases in the input-output
relations between these different parts of the system.

Although considerations such as these can account for many of the
mechanisms involved in a single stimulus-response sequence, they cannot
account for selective and integrative processes in more complex systems
which have a multiplicity of such subsystems in their repertoire. Here
the integrative and pattern-maintenance units, in their interaction with
the other two and with each other, are useful. The selection of cue mean-
ings in terms of their integrative significance in the system as a whole,
the storage and organization of information, and the selective legitima-
tion of responses all involve the functions of these units; in a very broad
way we have tried to show how the internal input-output processes can
account for these features of more complex behavioral systems.

Most generally we can say that the integrative unit consists in a
system of internalized object systems, in the more complex cases of a
social character. Their cathectic or motivational significance is primary
from this standpoint. Through the process of internalization their pre-
vious need-gratifying power as external objects has also become internal-
ized, so that there Is an organized internal reward system which operates
as the primary regulating mechanism for the adjustment of internal
strains and conflicts.

The pattern-maintenance unit, on the other hand, consists mainly
of internalized cultural pattern components in which values play the
central part, but it is also the primary repository of organized and codi-
fied knowledge. Thus, on the crude level of this discussion, we account
for the essential functions of memory and of internal normative controls.

Some Levels of Organization of Psychological Systems

Until this point, It has seemed advisable to speak of psychological
systems in general terms rather than attempt to discriminate different
types of systems on any basis other than that of the relative primacies of
functional components in the value systems and in Input-output inter-
changes. Effort to carry through our whole analysis for each of several

682 TALGOTT PARSONS

levels is not possible in the allotted space, yet confining it to one such
level, the personality, for example, would unduly restrict generality, the
demonstration of which is so important an aspect of our theoretical
intention.

Let us, therefore, attempt at least to put the problem into perspective
by briefly discriminating a few levels of organization. It should be clear
that this basis of classifying systems cuts across the functional types de-
fined in terms of value-component primacies which were outlined earlier.

It is possible to deal with a range which extends from the adult
human personality, treated as a total psychological system, to the simplest
S-R-S unit system involving a single simple behavioral goal, a simple
response pattern, and a very simple conditioned stimulus and drive
situation.

Let us start with an S-R-S unit. Most important for present purposes
is that, as Olds has demonstrated, it can fruitfully be treated as a psy-
chological system in the fullest sense of the theory of action. Olds treats
it as a system of three units (the number may or may not be arbitrary).
With a plurality of units (whatever their number) their interaction (on
the intervening variable level) is an essential focus of analysis for proc-
ess in the system. Secondly, in this case there is a particularly close inter-
penetration between psychological system and organism in that there is a
psychological unit for each discriminable component of observable phys-
ical behavior. This definitely cannot be said of higher-order psychological
systems, where the interpenetration must center on the central control
mechanisms of the organism, many of which probably do not have, in
the psychologically relevant processes, important physiological functions.53

Even in animal behavior, it seems certain that the personality is not
just an aggregate of elementary S-R-S units but that higher-level organ-
izations of the kinds Olds refers to as object systems and temporal systems
come into play. To take one example, we speak of a "hunger" motive.
But most animals will accept as food not just one class of physical ob-
jects but a variety of objects which resemble each other very little phys-
ically. When the animal is hungry, it does not seek only an object of a
particular physical class but "something to eat." Hence, there must be a
"food complex" which organizes several different classes by their func-
tional similarity, rendering them substitutable for each other within
limits. Therefore, the animal cannot react only to cues which indicate
the availability of particular physical goal objects, but must be able to
react to the availability of food as such. This category entails a fairly

63 Thus, it seems probable that highly abstract thinking, such as is involved
in mathematical reasoning, does not have any palpable physiological function for
the particular organism, e.g., cessation of it would not interfere with "health." But
it certainly involves complex neurological processes. Here the brain may be
functional to the personality rather than vice versa.

Psychological Theory in Terms of Theory of Action 683

high level of generalization since it has many diverse instances. For this
reason, we may suggest that the food complex constitutes an object sys-
tem organized about a physiologically given need but containing a com-
plex network of learned properties about food objects which has become
stored in organized form in the animal's "memory."

Similar considerations apply to what Olds calls "temporal systems/354
namely, complex chains of instrumental and goal-oriented relationships
between which there is often a relatively high degree of contingency, so
that with the failure of one set of expectations (for example the prey
gets away and the food quest must begin anew) another instrumental
sequence can be set into motion. This sensitivity to a range of alternatives
is one of the paramount characteristics of "intelligent" behavior.

We suggest that what Olds calls (internalized) object systems in the
primarily cognitive sense are located mainly in the pattern-maintenance
subsystem of the psychological system. The primary criterion of their
integration is the consistency of pattern elements with each other. Be-
haviorally, this consistency is observable in a low degree of contingency
in moving from one "aspect" of the system to another, and by the re-
versibility of relations. Temporal systems, on the other hand, are, as
internalized, located primarily in the integrative subsystem. They are the
internalized counterpart of the system's adaptability to changing circum-
stances, with a pluralism of instrumental alternatives and of particular-
ized goals integrated into a system.

As noted, it is a paramount feature of social objects that because of
the property of double contingency, they tend to share the characteristics
both of object systems in the narrower physical sense and of temporal
systems. This feature certainly is associated with the fact that their in-
ternalization constitutes a primary feature of the attainment of the higher
social cultural level of psychological organization. In other words, the
basis on which the motives for these rewards can be brought under con-
trol within a single organized system is the diffuse cathexis of a social
object, for the child ordinarily first the mother, which can include a
multiplicity of more specific rewards. Hence the very young, the oral,
child does not yet differentiate in its internalized organization between
these systems. We have suggested that only after the oedipal period
when the superego is formed is there a clear distinction between an ob-
ject system, the cultural norms of the superego, and a temporal system,
the internalized social objects of the ego.

We have said repeatedly that all systems of action seem to be organ-
ized in hierarchies of control. The above considerations suggest that in
psychological systems — and probably in others as well — there is a pat-
tern of alternating "layers" of system types in this hierarchy. One type
is exemplified by the elementary S-R-S system. Its primary characteristic

04 These seem to be very similar to Murray's "proceedings" as Olds recognizes.

684

TALGOTT PARSONS

is the functional predominance of adjustment of the system to the ex-
ternal situation through adaptive and goal-oriented activity. This type
may, following Olds, be called a "motivational system."55

The second type of system is organized primarily about the pattern-
maintenance and integrative units. Its focus is the system of internalized
objects in the widest sense, which includes both high- and low-contingency
types; we may refer to it as the internalized "object pattern" system.

Next it seems clear that these two types of system constitute alter-
nating layers within a larger system, e.g., a personality, which are not
concretely distinct, but interpenetrate. The principle of control hierarch-
ies makes clear the nature of their interpenetration : the A and G com-
ponents of a motivational system are articulated with the next lower
pair of L and / components in the series to form a system. This is to
say that at any given point in time the internalized objects of previous
phases of learning underlie and regulate the operation of motivational
systems. For object-pattern systems on the other hand, the relevant A
and G components are the ones "below" the relevant L and / com-
ponents in the series; i.e., the necessary situational adjustments to main-
tain an L-7 system are carried out through the next lower-order motiva-
tional-behavioral mechanism system. An example of such a subsystem
would be concern with relatively secondary "operational" reference
within a personality primarily concerned with abstract thinking (as in
the case of mathematical work) .

Given this mode of interpenetration, it is possible to arrange the sub-
systems of a more complex psychological system in a hierarchical order.
In our technical notation they appear as follows :

V

DIAGRAM 3
55 In the unpublished working paper referred to in note *, above.

Psychological Theory in Terms of Theory of Action 685

So far as we are able to ascertain, all the interchanges between functional
units thus discriminated are empirically significant.56

It should be noted that there are also "lateral33 relations of interpene-
tration and boundary interchange between subsystems. The most obvious
of these is the S-R-S or the c 'means-end chain/' in terms of which a
proximate goal state is an instrumentality for a further and more remote
goal, etc. The selective feature of the organization of such chains, which
crisscross each other in complicated ways, must, we hypothesize, be ac-
counted for by their regulation in terms of the L-I subsystems which lie
above and below them in the hierarchical order. Clearly also the L-I
components of subsystems are integrated together in larger integrates.

Finally, it may be noted that the principle on which the hierarchical
aspect of the structure of psychological systems is organized has much to
do with levels of generality in the cultural pattern content of the internal-
ized objects. The most familiar case is cognitive "control" of diverse
empirical data by bringing them under a generalized theory from which
particulars can be "derived." With proper consideration of the motiva-
tional component, we believe that this is a prototype of the organization
of systems of action.

Now let us take up very briefly another trend of differentiation of
the more macroscopic subsystems of psychological systems which is found
only in the human personality. We have already discussed what we be-
lieve to be the "tendency" for psychological systems to differentiate on
functional lines. We suggest that this occurs in the human personality in
a rather special sense because of the conditions of early socialization.

According to our model the socialization of the human child ideally
results first in the establishment of a personality system primarily organ-
ized about an internalized social object (the oral mother) and then
differentiated (on this level of organization, not the S-R-S or even
nonsocial object-system levels) into what may be called four primary
need-dispositions. The focus of this differentiation is the internalization
of the four primary role categories of the nuclear family, differentiated
from each other on the two axes of generation and sex. The child's own
role, of course, is ascribed in terms of his belonging to the filial genera-
tion, and to one or the other sex, but in his own personality all four basic
roles, i.e., son, daughter, father, mother, are internalized as social objects
and constitute the main scaffolding of the personality system at this stage.

As applying to the oedipal personality we have referred to these need-
dispositions as (1) adequacy (internalized son-brother), (2) security

BC This statement has been tested through a good deal of substantive work deal-
ing with the "stratification" of the personality as resulting from the phases of the
socialization process. It has not, however, as yet proceeded far enough to justify
reporting on here.

686

TALCOTT PARSONS

(internalized daughter-sister), (3) nurturance (internalized mother), and
(4) conformity (internalized father) [cf. 26, chap. 2]. Ego's self is cate-
gorized of course in terms of either the son-brother or the daughter-
sister role. But the difference is a matter of primacy, and all four of the
objects are internalized in personalities of both sexes.

These four primary need-dispositions originate in role behavior within
the nuclear family. With the child's post-oedipal "emancipation" from
the nuclear family the requisite objects are "lost," but as the scaffolding
of personality structure, they persist throughout life. This suggests that
raising the child, through oral dependency, first to the level of an in-
ternalized social object system, and then differentiating this object system
along primary functional lines, provides the indispensable basis for a
personality which can give primacy to cultural-level organization of be-
havior. Whatever the limitations in the neurological structure of sub-
human species, it seems clear that the absence (or rudimentary char-
acter) of socialization experience gives such primacy to the adaptive-
gratificatory exigencies determined by the nonsocial environment that
it is never possible to attain a human level of organization. The animal
must give too heavy primacy to "motivational" systems; he has no
chance to develop sufficiently complex and highly organized object-
pattern systems.57

After the establishment of the primary need-disposition structure,
which persists only in modified form, the child must develop a new and
crosscutting type of psychological system; this may be called a role-
orientation system. Of course, the need-dispositions originate as role-
orientation systems, but these roles are superseded. What I am discussing
now is the role structure of the adult personality.

The first point to be made about adult role-orientation subsystems,
e.g., the occupational role, is that they involve all the primary need-
dispositions. The primary need-dispositions are learned subsystems of the
personality, the goals of which are not constitutionally given — except in
the sense of some sort of "potentiality." The role-orientation systems are
learned at a later time on a crisscrossing basis; hence they are removed
from a constitutional base by at least two major steps not merely of
elementary learning but of profound systemic reorganization of the whole
personality system. Any attempt to treat the "motivation" of activity in
such roles as a simple manifestation of a "propensity of human nature,59
meaning by that a constitutionally given "drive" or "instinct,," is grossly
unsatisfactory. For not one but several layers of internalized objects
operate between the constitution of the organism and the control of

BT Because of limitation of space this sketch is extremely schematic. Much
fuller discussion will be found in [25, Chaps. 2, 3, 7] and in "Social Structure
and the Development of Personality," op. cit.

Psychological Theory in Terms of Theory of Action 687

overt behavior. Motives themselves, and the sanctions to which they are
sensitive both externally, from alters in the situation, and internally,
from other subsystems of the personality, possess a degree of generality
in meaning which cannot in any sense be found in instincts. Neither
can such motivation complexes be interpreted as mere aggregates of
conditioned reflexes or even of "habits," in the atomistic S-R-S sense.
They are highly organized systems.

Let us illustrate briefly with one example, the motivation of adult
occupational activity. This is simultaneously a role in a nesting series of
several different social systems. In the typical modern case, the focal
one is the organization in which the individual is employed, e.g., a busi-
ness firm, a government agency, or a university, which in turn serves
functions in larger social systems. By virtue of his contract of employ-
ment, ego is obligated to perform a whole complex of services to the
organization, often breaking down into very diverse subcategories with,
however, some kind of primacy relationship. Thus, for a university
faculty member teaching and research — themselves very complex activi-
ties— usually have first place, but he is also a member of the faculty and
the department and has some responsibilities for their conduct as organ-
ized bodies.

A crucial characteristic of most occupations is that the "product" is
not typically utilized directly by the producer: the teacher only sec-
ondarily teaches himself; the shoe worker wears very few of the shoes
he helps to produce. Hence there must be some other motivational mech-
anism than in simpler cases is provided by the immediate goal value of
the outcome of his activities. To account for this we must distinguish
external rewards dependent on the situation — typically sanctions — from
internal rewards. In the first class there are typically two essential sub-
categories. One is "remuneration," e.g., money as a generalized medium
of exchange with which particular goods and services can be acquired.
Remuneration establishes a relation to "intrinsically" significant goal
objects, but through a highly generalized and peculiar mechanism very
different from what is found in most rat-running experiments. Other
"tangible advantages" may also be gained, such as prospects of promo-
tion to more remunerative jobs in the future, various kinds of power or
influence, etc. The second class of sanction rewards may be called sym-
bolic and attitudinal; these are valued not for what they "do for" the
actor as goal objects but for "what they mean." Money remuneration
itself has a symbolic aspect as a measure of the regard in which a person
is held, of his "success." But beyond this, various kinds of approval,
recognition, honor, etc., must also be taken into account.

Turning to the internal rewards, the most essential point is the rele-
vance of internalized values to the motivation of such activity. It seems

688

TALGOTT PARSONS

to be well established that the motivation to types of occupational achieve-
ment which involve highly intangible, not in the cruder senses "consum-
able," results cannot be understood without reference to the internaliza-
tion of values of work and of achievement through the formation of ego
ideals and superegos.58

The reward system for sophisticated human achievements must quite
clearly be organized on highly generalized cultural and symbolic levels.
We do not yet have an adequate technical analysis of these things, or a
terminology, to use in discussing them. But to revert to common-sense
terminology, terms like achievement, success, wealth, power, fame are
indispensable, as are, on the internal side, concepts like self-respect. A
psychological theory which cannot interpret the common-sense meanings
of concepts like these except to "debunk" them cannot hope to approach
the analysis of the mature human personality and its linkages with the
social and cultural systems.

Structural Change in Psychological Systems

Whether inevitably or no, the type of theoretical analysis set forth
in this essay seems naturally to start with the presentation of the struc-
ture of a system and the processes by which a system with that kind of
structure "functions" in relation to its environment. But "structures" do
not remain unchanged. They come into being through some kind of
developmental process, and they pass away and are replaced by other
structures. Our approach in no way implies that structure or equilibra-
tion is empirically more important than change. But it will help give per-
spective to discuss the approach of a theory of action to problems of
structural change.

First, on the most general level of theory there is no difference be-
tween a theory of the equilibrating processes of systems and a theory of
their processes of change. A theoretical system in the analytical sense is
not a set of empirical generalizations; it is a system of concepts and their
logical interrelations in generalized propositions at various levels. "The-
ories" of equilibrium and theories of change are applications of general-
ized analytical theory. The differences between them are parametric,
not theoretical.

The crucial parametric differences concern the stability of the "struc-
ture of the system," which we have associated with the internalized value

53 On the psychological aspects of this compare [19] especially. On the socio-
logical aspect a particularly famous case study is that of Max Weber, The Protes-
tant Ethic and the Spirit of Capitalism. I have concerned myself with these prob-
lems from time to time over a long period. For a relatively early discussion see
"The Motivation of Economic Activities" [23]. The most recent discussion will be
found in [27, Chap. 3].

Psychological Theory in Terms of Theory of Action 689

system. This stability is a function of various other features of the internal
character of the system, e.g., its level of integration. It is also very partic-
ularly a function of the relation between the system and the ranges of
variation of its significant situational conditions. Every empirical con-
ception of equilibrium must include some conception of the limits with
respect to the essential inputs which can be tolerated by the system. For
example, an organism has definite limits with respect to nutritional in-
take, oxygen, etc.

Three things can happen to a system, including a psychological sys-
tem, when the input-output factors approach "dangerous" limits in
relation to its environment. First is the manifestation of "strain," of "ab-
normal" functioning of some sort. This is always the initial consequence,
and may become consolidated into a "pathological structure," i.e., one
which is relatively stabilized in such a way as to cope with the specific
sources of strain, but at some "cost" elsewhere.

The two other possibilities are more fundamental. One is dissolution
of the system. For the organism this is death, but social systems can dis-
solve without either the physical or the psychological death of their mem-
bers. Another is structural change which makes possible successful adap-
tation to environmental situations outside the previous limits of tolera-
tion. The three possibilities are dynamically related. In the process of
change there is always "pathology"; there is always "a little" death in
the sense that some old structures are eliminated. Short of complete dis-
solution there is always some structural change; it is difficult to imagine
"pure" pathology over a long period without any structural change.

To us, learning and structural change in psychological systems are
identical. In interpreting this statement, it is essential to be clear about
the level of system definition and organization to which it applies. The
human personality is continually learning new things, but this does not
necessarily constitute in a relevant sense structural change in the per-
sonality as a system, but only in very subsidiary subsystems at a level re-
moved by a good many steps from the personality as a whole. We can
speak of someone as having "a very stable character" through a great
deal of such change. At the opposite extreme, the basic stages in the psy-
chosexual development of the individual, e.g., the oral, oedipal, and
adolescent stages, mark fundamental structural changes in the personality
as a system. These, too, are in the most general sense "learning proc-
esses," although it does not follow that what is usually called "learning
theory" is necessarily adequate for their understanding.

In a very tentative way, we have sketched a model of the process of
adaptive structural change in a psychological system; we believe that
this model is of very general significance in the theory of action, at sev-
eral psychological levels and in social and cultural systems as well. It will

690 TALGOTT PARSONS

be remembered that the model involves frustration of important goal
interests, expression of the consequences of disturbance, diffuse sup-
portive input from the environment, provision of new facilities, selective
rewarding of new performances, and reinforcement by a systematically
structured pattern of sanctions.59

Work with this paradigm has not yet proceeded far enough to per-
mit a competent judgment of its accuracy or of its generality over any
wide range. It was to be expected, however, that in the general theory of
action there should emerge a paradigm of structural change of a level
of generality comparable to the phase pattern of performance processes
worked out by Bales and others for the small group. Olds has shown
how that paradigm applies broadly to elementary psychological systems.
It also broadly fits the pattern of the trade cycle in economics. On the-
oretical grounds we argue that in a structural change cycle the order of
phases should, from a certain point of view, be the reverse of that char-
acteristic of a performance or equilibrating cycle. On the level on which
we have worked, this turns out to be so.

Part III

SOME QUESTIONS OF METHODOLOGY AND OF THE

SCIENTIFIC SIGNIFICANCE OF THE SYSTEM

In this concluding section, I shall try to bring together very briefly
discussions of the topics formulated under rubrics 13} to {Yi\ of the out-
line. Some of them have been dealt with earlier and in these cases I
shall avoid repetition as much as possible.

Initial evidential grounds for assumptions of the system. Part of the
answer to this question has been anticipated under Background Factors
and Orienting Attitudes. In the particular case of this scheme it is nec-
essary to distinguish evidential grounds for the general theory of action
from those for its treatment of psychological systems. As has been noted,
the most salient evidence for me was, in the first instance, sociological.
Above all it came from the field of comparative study of institutional

59 Cf. above. The model was presented by Parsons and Olds in [25, Chap. 4]
and generalized by Parsons and Bales in chap. 7. In that context it dealt primarily
with a single cycle in the process of socialization, but with the appropriate para-
metric modifications, we believe it would fit other levels of psychological change.
One social system case has been worked out, in application to a structural change
in the economy, illustrated by the separation of ownership and control in the
American economy during the past fifty years. This is published in [27, Chap. 5]
and also in Explorations in Entrepreneurial History [28] in a paper by Parsons
and Smelser. Smelscr is now engaged in testing it in a much more complex
empirical case, the development of the British cotton textile industry, 1780-1840.

Psychological Theory in Terms of Theory of Action 691

structures as undertaken by Max Weber, whose particular focus was on
the relativity of the institutions and behavior patterns associated with
the modern Western economic order. This concern led into psychologi-
cal variables, since the economic doctrine of "self-interest," the point of
departure of my technical concern with psychological theory, is partly
based on psychological assumptions. In the theoretical categorization of
the evidence concerning institutional structures, my starting point wa?
the problem of classification of social roles, as they differ both among
societies and among role-types in the same society. Above all, it was in
defining the contrast between business and professional roles that the
elements of the scheme of "pattern variables" took shape. (See above,
under the discussion of Structure of the Theoretical System.) This
scheme is relevant to a vast amount of empirical data in sociology.

On the psychological level, the main evidence comes from data on
the motivation of role behavior and the psychological bases of behavior
patterns manifested in different roles. Thus the facts mobilized from
several directions by Freud, with relation to the superego, by G. H.
Mead, with relation to "taking the role of the other" and by Durkheim
with special reference to anomie constitute reference points of primary
importance. Data concerning the internalization of values and of social
objects as psychological phenomena first gave strong impetus to my
theoretical thinking on psychological matters. Several other bodies of
evidence, such as Weber's account of how religious ideas could influence
behavior, seem to be congruent with these.

From this point my interest ramified into problems concerning the
structure of personality, with reference to levels such as those dealt with
by Murray in his analysis of needs, and into the genesis of personality
in the process of socialization, where the facts mobilized by Freud and
other psychoanalytically oriented writers., particularly perhaps Erikson,
were salient. Only much later did specific empirical evidence from the
more microscopic levels of psychological analysis, as distinguished from
general interpretive views, become particularly important for my own
theoretical development. Here my principal interest has been in codifi-
cation, namely, looking for a fit between available evidence and the re-
quirements of the general theory of action.

It is perhaps understandable that, for a sociologist in the initial
stages of psychological interest, particularly concerned with relatively
macroscopic sociological problems, the "data," i.e., the assertions of fact
from psychology which are most relevant, have been of a relatively gen-
eral character. To me, it is a central methodological point that a fact is,
as noted earlier under Orienting Attitudes, a statement or proposition
which has been empirically verified. It is stated in terms of a conceptual
scheme and may be couched at any level of generality; of course, the

692

TALGOTT PARSONS

more general the statement the more difficult it may be to verify, though
this is not always the case. Hence a statement of Freud that the super-
ego "represents/5 i.e., internalizes, the parental function in the family
of orientation I consider a statement of fact. The important point is that
the "immediate data55 level, and the appropriate "language" for de-
scribing such data, is relative to the scientific problem in hand. To the
sociologist interested in certain psychological problems, a generalization
of Freud's may be an immediate datum, though to Freud himself only
materials derived from the observation of particular cases would con-
stitute such data.

Construction of function forms. Three principal topics seem to be-
long under this heading: the problem of intervening variables, that of
prediction, and that of models. The logic of the independent-intervening-
dependent variable scheme is, of course, central to the structure of the
outline. In this regard, the most important single point I wish to make
is that it lies on a different, more empirical level from that of the main
structure of the theory of action and its psychological subsystem which
has been set forth earlier. The essential difference is that the independent-
intervening-dependent variable scheme is concerned with the interpreta-
tion or prediction of a particular empirical event or class of such events.
The general theory, on the other hand, is not restricted in this way, but
is applicable to any empirical phenomena falling within its range. When,
however, a specific empirical problem is tackled by the general theory,
it can be cast in terms of the other scheme.

Ideally, the best procedure would be to deal with all the funda-
mental variables of the system simultaneously for the solution of any
given empirical problem. Even in such a case, however, many empirically
relevant considerations would have to be treated as given parameters
of the problem. In other words, only one subsystem of the theory of ac-
tion (or one boundary-interchange process between two subsystems)
could probably be formally and explicitly handled in a single analytical
procedure. If, therefore, the problem in hand were psychological, the
organic, social, and cultural factors would have to be treated as given.
Presumably the same is true of the main facts about psychological sub-
systems of the total personality other than the one directly under analysis.
This, however, is by no means to say that the empirical systems appro-
priate to analysis in each of these theoretical terms must be treated as
unrelated, a set of watertight compartments of the empirical world. On
the contrary, they are most closely interdependent, and we have shown
that specific analyses of the relations between them (e.g., between per-
sonality and social system in the area of "object-relations") are of the
first importance and are theoretically quite feasible. The question is how
such intersystem relations can be technically analyzed. The answer

Psychological Theory in Terms of Theory of Action 693

seems to be that the processes of a given boundary interchange can them-
selves be treated as a system of action, using the same basic paradigm
which has been used for intrasystem processes.60

The kinds of restrictions on the generality of theoretical analysis
which have just been noted unfortunately do not give the whole story.
We are not yet in a position to treat all the variables of a system of ac-
tion in terms of a set of simultaneous equations. Hence for the empirical
solutions which we can reach in the present state of knowledge, it is
necessary to select certain of these variables as independent and to hold
the others constant for purposes of the particular problem, or at best,
to introduce ad hoc qualifications of the implications of the assumption
of constancy which we can derive from our incomplete knowledge of the
interrelations of the selected variables with the others. Thus we are
forced to use the independent-intervening-dependent variable scheme,
but this is not the logical ideal except for handling particular empirical
problems.

A second question is what variables are "intervening." The essential
point here is that very likely most, if not all, of the variables employed
in the theory of action are intervening variables for some types of sys-
tems, but these are not the same for all cases. To take one example, for
many types of personality analysis, such as that attempted above, it
seems that most of the processes of boundary interchange or communi-
cation between units internal to the system are not directly observable
and hence must be treated as intervening variables. As independent and
dependent variables, the observables must be found in behavior, which
may by symbolic, including verbal. But the case is different for the social
group. In the type of experimental study of small groups carried on by
Bales and his associates (and others) the unit of the system is the mem-
ber-in-role. Processes of communication between these member units
constitute the principal empirical subject matter of direct observation.

This kind of difference provides an important source of strength for
such a scheme as the theory of action. To be sure interpersonal communi-
cation in group behavior is not the same thing, empirically, as intra-
psychic communication between units of the personality system. But if
we are right that the same variables are involved, relations between the
variables which are established from the study of interpersonal com-
munication should, if sufficiently generalized, prove applicable to the
processes of intrapsychic communication. It is necessary to be extremely
careful to determine the parametric differences correctly, but this is not
by any means inherently impossible. And I think it is very unlikely that

co This has been attempted, within the general social system framework, in
[27, Chap. 3] for different subsystems of the society. Its extension to the inter-
changes of social and psychological systems seems altogether feasible.

694 TALGOTT PARSONS

any of the essential variables of the general theory of action must be
treated as intervening variables for all types of system to which the
theory is relevant.

The problem of prediction is very closely related to that of the use of
the paradigm of independent-intervening-dependent variables. This is,
to my mind, essentially a paradigm for formulating either a prediction
( a hypothesis of course is an application of the logic of prediction ) or an
interpretation of a past event, which is logically speaking a retrospective
prediction. Thus I take for granted that one of the primary tests of a
theoretical scheme is its usefulness in prediction. But the generality of
this usefulness is rooted in the levels of theoretical systematization which
have been the main subject of this essay. If theoretically based, successful
prediction is an application of general theory. But it is not the only
such application; codification of factual knowledge is another con-
spicuous one, an important example of which is the systematization of
morphological classifications, so important in the biological sciences.

Predictability is, furthermore, a valid test only under conditions
which must, methodologically, be very carefully defined. To take one
common example, the difficulties of predicting day after tomorrow's
specific weather do not constitute an adequate test of the validity of the
systematic theory of physics in the field of the behavior of gases.

It is an implication of the Whiteheadian view I have stated earlier
that there is a presumption that, in any concrete set of phenomena, a
plurality of theoretically defined systems may intersect. Any one such
system can then serve as a basis of accurate and detailed prediction only
when the data conform to certain specific methodological requirements.
Whether or not these are controllable in practice, they must be the "ex-
perimental conditions" for observing the generalized uniformities formu-
lated in the theory. When empirically such conditions are not given, then
two recourses are open. First, it may be possible to attempt to achieve
relatively precise predictability by treating specific phenomena as re-
sultants of the operation of two or more theoretical principles, perhaps
deriving from different theoretical systems. Second, one may use various
techniques to discriminate between the effect of the variables relative
to one level of theoretical analysis and some order of residual variance
which can be more or less randomized for purposes of the particular
analysis and tested for by statistical techniques.

One particular caution is necessary when the adjective "psycho-
logical" is applied to prediction. We have noted that any theoretically
based prediction must be hedged by the possibility that variables or fac-
tors other than those included in the system being used for prediction
may influence the result. As systems of behavior, psychological systems
may be influenced by the independent operation of processes in physio-

Psychological Theory in Terms of Theory of Action 695

logical and biochemical systems, and the usual methodological precau-
tions would have to be taken to isolate the psychological system from
these.

But this is not all. If we are correct in our view that psychological
systems constitute one special class of systems defined in terms of the
broader theory of action and that in general the other classes are in-
volved in all concrete behavior, especially on human levels, then there is
never a presumption that concrete behavioral processes can be ade-
quately explained or predicted from purely psychological analysis unless
the factors appropriate to social and cultural systems, as well as physio-
logical factors, are adequately controlled. Furthermore the personality
is a complex organized system with many subsystems differentiated from
each other on various levels. Hence careful discrimination of type of sys-
tem and of system level within the category of psychological systems, as
well as discrimination of psychological from other systems, is necessary
for valid prediction based on psychological theory.

I have no particularly clear set of views about the utility and role of
models. The most general sense of the term model seems to be that of an
"ideal type" of structure or process, arrived at by hypothetical reasoning
from theoretical premises, which is then used, through comparison with
empirical data, to analyze such data. In this meaning, model seems to be
almost identical with theoretical scheme. And, if it is theory, and of
course, good theory, I am in favor of it.

Apparently, the term model is used for conceptual schemes on many
different levels of generality. For example, Tolnian's "A Psychological
Model" [22] comes very close to being a general statement of his theo-
retical position in psychology. Models seem to vary all the way from
this level to formulations of very specific processes.

I wish to call attention to one danger in the use of models which
has come into focus in connection with some work on the border line be-
tween economics and sociology [cf. 27]. This is the tendency to isolate
a specific set of independent variables within a parametric setting, and
then refuse to say anything about anything else except that the param-
eters must be treated as "given data." This is justified for certain
limited purposes, but it often tends to engender a habit of mind by which
an artificially sharp line is drawn between the problem area on which
attention Is focused, and related areas which both empirically and theo-
retically impinge upon it. It favors a special case of the fallacy of mis-
placed concreteness, in that it is easy to slip from the position of holding
the factors involved in these data constant for particular methodological
purposes,, to the implicit assumption that they vary, if at all, only in
ways which do not affect the internal structure of the model; i.e., their
variations are assumed, usually implicitly, to be random.

696 TALGOTT PARSONS

Two striking examples of this tendency may be noted. First, al-
though the more sophisticated modern economists have been careful to
hold to the given data formula about many borderline problems of eco-
nomics, often implicitly and sometimes explicitly (cf. the work of Lionel
Robbins) they have tended to say that it is a condition of economic
analysis that it should operate in a setting where there is no theoretically
intelligible order on the border line of economic processes; i.e., the data
are not merely given, but random. A second example, this time from
psychology, lies in the fact that in much learning theory work, particu-
larly that stemming from Hull, no specific analytical attention is paid
to the structure of the stimulus situation; it is "given," usually set up by
the experimenter in a form unalterable by the rat. The paradigm has
been that of an animal pursuing a given goal, e.g., hunger-drive reduc-
tion in a given situation. It is never explicitly denied that, for example,
human children are placed in complex situations of interaction with
others (children and adults) where the structure of the situation is any-
thing but given. Yet when Hullians have started to generalize beyond
their favorite experimental situations, they have argued as though the
variant structure of the situation did not matter.

I am quite aware that these dangers are not inevitably connected
with "model-theorizing35 and that neither all economists nor all Yale
psychologists have, in this sense, misused their models. But this has
happened often enough to make a call for caution appropriate. The im-
portant point is that every model must be conceived to have a theoreti-
cally ordered context; it is a product of the special abstraction of part of
a range of theoretical relationships for specific purposes and must never
be "reified."

By and large, the idea of model is on the periphery rather than in
the center of my thinking about theory, though I have occasionally used
it [e.g., 27, chap. 5]. For me, the idea of systematic theory is the central
one and models are as likely to be harmful as useful if they are not
carefully related to theoretical systems.

Mensuratlonal and quantificational problems. On this point I take
an essentially pragmatic position. I do not depreciate the importance
of measurement or of the quantitative treatment of data secured by
measurement. Their importance stands out as overwhelmingly great in
almost all branches of science. At the same time, I definitely refuse to
concede that knowledge which is not put in these terms is always "both
meager and unsatisfactory."61 Such a doctrine rules out a very large
part of biological science and in our discipline much of the use of clini-

61 The well-known statement of Lord Kelvin used as a motto for the Social
Science Research Building at the University of Chicago.

Psychological Theory in Terms of Theory of Action 697

cal observation and anthropological field work. I regard the common
dogma that quantitative techniques alone have scientific value as similar
to the old behavioristic dogma that data of verbal behavior were scien-
tifically inadmissible because they were "subjective"; the quantification
dogma seems destined for a similar fate.

A prime theoretical consideration here is that quantification makes
sense only when concepts and their logical relations are well defined.
For reasons inherent in the structure of logical systems, these definitions
can never be exclusively quantitative; if they were, the state of any
empirical system, however complex, could for every purpose be repre-
sented by a single figure. There is a "qualitative" component in any
theoretical structure, which must be reflected in the categories in which
data are classified and classes are related to each other. Similarly I
would radically deny the dogma, popular in the early phase of the vogue
of operationalism, that operations of measurement can, in their scientific
significance, be reduced exclusively to the physical manipulations from
which numbers emerge, with the implication that all theory is purely
inductive generalization from the results the operations produce.

The first essential point is that quantification works from a qualita-
tive base in the logical structure of theory; the latter can never be
eliminated, though of course it changes. Second, this qualitative com-
ponent of the theoretical base plays an essential part in determining ob-
servational (including measurement) operations; questions of the signifi-
cance of the results of measurement, if not the concrete data themselves,
never can shake free of this logical dependence on qualitative considera-
tions.

Within this methodological framework, the degree of specificity of
measurement and quantification is not a question of methodological
principle but of their relative fruitfulness for particular purposes. This
in turn will depend on a number of factors, such as the kind of data ac-
tually and prospectively available, the techniques of measurement avail-
able to secure them, and the relations of these techniques to other ob-
servational procedures, the kinds of analytical and processing proce-
dures available for handling data, and above all, the state of the theo-
retical scheme and its relation to these other problems, which will de-
termine what order of theoretical significance can be attributed to what-
ever quantitative generalizations can be formulated and empirically
validated. All this makes it a very complex problem area.

I believe that there is, in a sound development of science, a general
trend toward greater quantification, but that this is likely to be maxi-
mally fruitful, i.e., theoretically significant, after the requisite spade-
work on qualitative levels when there has evolved a well-articulated
analytical scheme in terms of which operational problems of measure-

698 TALGOTT PARSONS

ment, notably defining what should be measured, can be stated. How-
ever, this is a matter of relative predominance and not of exclusive
either-or. Both types of methodology will be found at all stages but with
increasing prominence of the quantitative.

Furthermore, it has often been pointed out that there are circum-
stances where too great empirical precision (quantification is a form of
precision) may be a positive hindrance to scientific advance. This is
because such precision may distract attention from broader relation-
ships, concentrating it on detail. Thus, had Newton had the data avail-
able to Einstein, but the mathematical-theoretical equipment he actually
had, he well might never have reached any major synthesis, because the
"Newtonian" theory would have been empirically inadequate, and he
could not have attained the Einsteinian. I believe that this is a really
serious problem in the action field. Over considerable ranges of it there
is a sense in which we know too much; a very great problem is how to
select among the available data in terms of their theoretical significance,
as opposed to exploring the "fine" quantitative relationships which are
still unknown in any sense [cf. 20].

I am also inclined to think there are prejudices to the effect that any
theory which is at all general should be capable of being expressed in
terms of some specific type of mathematics. The great model of course
has been the differential calculus as the form used in Newtonian mechan-
ics. This has played an enormous part in the history of science, but
there may well have been an element of chance in its magnificent sci-
entific utility. The logical tools now available are far richer than in New-
ton's time, and many different emphases are possible.

On the whole, I believe that the behavioral sciences are following a
course of development not unlike that of the biological sciences but are
now still in an earlier stage. Only relatively recently has quantification
(and certainly higher mathematical analysis) begun to play an ex-
tremely prominent part in biology. Yet it is most emphatically false that
all the preceding qualitative work on the classification of species, on
anatomy and histology, on "either-or" experiments, etc., has been in
vain. Even in physiology and biochemistry an immense amount of the
most significant work has been qualitative, e.g., simply identifying by
qualitative chemical analysis the specific chemical compounds present
in different physiological processes, and then perhaps resorting to the
crudest possible quantification, e.g., in terms of the consequences of
presence vs. absence.

I believe that tbfe biologists' task has been easier than ours because
of the immense range of organic species available for observation and
comparison, and the clear empirical differences of their crudely ob-

Psychological Theory in Terms of Theory of Action 699

servable properties, as well as because of the greater opportunity for
controlled experiment. There is a sense in which the basic classificatory
schemes of biology have emerged from relatively "obvious" groupings
of data, in a sense in which this has not been possible for the behavioral
sciences.

Thus very broadly, I think that the theory of action is in a state not
very far from but in several respects still behind that of biological theory,
at about the stage when really "modern" physiology appeared. The pri-
mary systematization we have so far achieved deals with "analytically
descriptive" classifications of structures — structures of roles, collectivities,
and institutions, or of need systems, which are often complex hierarchies
of structures. Beyond this, some "structural-functional55 analysis of proc-
esses within and between such structures in terms of the "mechanisms"
by which certain functional "needs" of such systems are met has proved
possible. The importance of this level of systematization should not be
underestimated. Among other things, it has provided a framework
within which the theoretical significance of more detailed empirical
problems could be formulated and a framework for codification whereby
structures and processes of different types of systems could be rendered
comparable.

There is, of course, an immense amount of quantitative information
available in our fields, but it is not yet very highly codified or adequately
related to larger theoretical analyses. This as much as anything delays the
much-desired cumulative development of knowledge, since cumulation
in the scientific sense is by no means a simple matter of increase of num-
bers of facts known; it involves organization of the facts, which can only
take place through generalized theory.

The methodological ideal is that the specific data obtained by ob-
servational procedures should turn out to be the values of the variables
of a generalized theoretical system; this was the case for part of the
astronomical phase of classical mechanics and has been true m a few
other cases. As yet, this ideal cannot be said to have been attained any-
where in the behavioral sciences. Economics alone has, for any length of
time, possessed a theoretical system of the requisite logical integration and
refinement, but repeated attempts to gather statistical data which pro-
vided the direct measures of the fundamental variables have failed. The
basic reason seems to be clear; the "experimental conditions" necessary
for success do not occur in "nature" any more than do certain of the
chemical elements.

On the other hand, where generalizations based on quantitative data
have been achieved and validated, so far no one has been able to fit
them directly into any generalized theory. Probably this is because what

700 TALGOTT PARSONS

has been measured is some order of resultant phenomenon, involving
usually not only several variables of the same system, but even more
likely the intersections of plural theoretical systems. A demographic gen-
eralization such as Pearl's logistic curve or Stouffer's treatment of inter-
vening opportunities would be cases in point [cf. 29, 33].

The version of the general theory of action presented here has taken
a turn which may prove to be the harbinger of quantifications closer to
the level of generalized theoretical significance than before. This is the
statement, as attempted above, of the equilibrium conditions of systems
in terms of the balances of inputs and outputs in relatively clearly defined
categories over the boundaries of the system. In a logical sense, this is
an inherently quantitative approach, since it involves for empirical ap-
plication the distinction between an optimum input or output and a too
much or too little which deviates from the optimum in one direction or
the other.

On very general levels, this input-output analysis has already been
applied to the problems of small-group equilibrium, to the genesis of
deviant behavior in social systems, to phases of the socialization process,
to the boundary processes of the economy, and in a highly tentative
way above to the processes of psychological systems.

Formal organization of the system. Again, my attitude on this ques-
tion is relatively pragmatic. One should distinguish between methods of
actual substantive theoretical work and certain ways of stating results
and subjecting them to certain types of test.

Methods of substantive theoretical work are complex and difficult
to describe. I and my associates have generally focused on empirical-
theoretical problems and problem areas, that is, problems which have
both empirical and theoretical aspects. The empirical elements have ap-
peared at several different levels both on the microscopic-macroscopic
range and with respect to the character of the relevant data. Thus, to
take the critical example discussed before, the problem of the nature
and limitations of the economic conception of "self-interest" was studied
in relation to the contrasting broad patterns of institutionalized role
behavior in the business world and in that of the professions, with a
range of comparative institutional structures outside our own society in
the background. Theoretically, the study dealt with the relation between
psychological and sociological problem statements, but on the whole it
worked from the sociological into the psychological rather than vice
versa. Another much more recent case would be the consideration of the
learned and institutionalized elements of sex role, with facts drawn both
from cross-cultural material on kinship and sex role outside the kinship
realm and from the socialization process in the family. Again, the theo-
retical interest has been both psychological and sociological [cf. 22, 25].

Psychological Theory in Terms of Theory of Action 701

Perhaps the most important generalization about our methods of
work is emphasis upon the immense importance of theoretical work as
such. A prejudice widely current in the culture of our branches of sci-
ence asserts that all the "hard work" of science lies in the empirical fact-
finding and validating process; once the facts are known their theoretical
significance and implications will automatically spring to mind. This, I
feel, is a very serious error. After all Galileo knew all the crucial facts
needed to explain the limitations of operation of a suction pump in
terms of atmospheric pressure, and he missed the explanation [cf. 6]. Ob-
viously in matters of scientific theory, Galileo was not a stupid man.
To see the significance of facts it may be necessary to have attained a
theoretical perspective whereby this significance acquires saliency; and
this is not likely to occur without effort.

Our theoretical work consists essentially in a laborious process of
trying out possibilities of combinations of the available logical com-
ponents of a theoretical scheme. It also investigates possibilities of modi-
fying both these components and their relations, in the form, for instance,
of the redefinition of fundamental concepts and the introduction of new
components. The case of the concept of value cited in the introduction
and used later is a conspicuous example.

For this purpose, various kinds of devices for formalization become
very useful tools. On the level on which I have been working, "equa-
tions" are still too specific, and I have used what, logically, are cruder
and more elementary procedures, especially relatively formalized classi-
fications. There is always the danger of hypostatizing such tentative
formalized statements, and some critics contend that the habit of using
them necessarily introduces fatal rigidities into one's thinking. Yet, what-
ever the dangers, I do not see how the levels of theoretical specificity and
generality we are aiming at and working with can be handled if in prin-
ciple we confine ourselves to discursive exposition.

For these classificatory purposes, the fundamental starting point is
the "fourfold table," namely, the simple cross classification of what in
some sense are the "polar" or widely separated values of two variables.
The general justification of using the logic of this procedure is related to
the extremely wide usefulness of binary discriminations: the binomial
theorem in mathematics, the "bit" theory in communication theory,
various trends in symbolic logic, and substantively, empirical evidence
such as the predominance of division into two in the biological processes
of cell division.

Turning from the field of workaday procedures of theory construc-
tion, formalized deductive procedure acquires great significance as a
test of validity and as a revealer of hidden premises and problems, at
certain junctures in the development of a theoretical scheme. In my

702 TALCOTT PARSONS

opinion, any working theorist who confined himself to this type of pro-
cedure would soon cease to be a theorist and become either a logician
or a philosopher. Nevertheless, as a completed structure, a theoretical
system should be capable of statement in logico-deductive form, starting
with axioms or postulates, and proceeding to definitions of variables,
theorems, parametric constants, deductions from theorems, etc.

For the theory of action, an attempt on this level was made in
"Values, Motives and Systems of Action" [22] ; in retrospect, this seems
to have been successful. It was far from a "definitive55 statement; indeed,
it revealed difficulties which have already led to substantial revisions.
The attempt was distinctly useful, however, and showed, for the first
time on a comparable level, the possibilities of strictly deductive develop-
ment of such a general scheme. Many of its elements, notably the re-
lational character of all action and the different system-formation refer-
ence points, I continue to adhere to; other features have been modified
and are expected to be modified further.

In sum, the logico-deductive form is ideal for presenting a theoreti-
cal system when it is relatively complete. Attempts to achieve this form
constitute one fundamentally important type of test of the state of a
theory, along with the empirical test, the most important one. From
time to time, it should, therefore, be seriously attempted. Yet deductive
procedure in this sense is not the primary everyday method of the work-
ing theorist; he is likely to be spending most of his time on relatively
specific problem-oriented levels, and thus like the empirical researcher,
not worrying about the ultimate validity of what he is doing.

Finally, I do not think any logico-deductive statement of any theory
is ever definitive in an absolute sense. As Whitehead made crystal clear
with respect to the classical mechanics, when it came to be stated in
sharply deductive form, it bristled with difficulties of which philosophers
were well aware long before scientific advance began to deal with them
effectively (e.g., the problem of "action at a distance" and the unre-
solved conflict between wave and corpuscular theories of light). The
theory of action certainly involves many philosophical difficulties. It
stands or falls, however, not on its meeting these difficulties, but on the
contribution it makes to empirical science, relative and approximate as
that always is.

Scope or range of application of the system. In view of the nature
of the system the problem of scope should be put in two different ways :
the scope of the general theory of action; and the scope of the theory of
psychological systems which can be considered to be a subsystem of that
general theory. I see no answer to the first question other than that the
theory must be held to cover the whole range of the "sciences of be-
havior" as ordinarily defined or what have been a little more narrowly

Psychological Theory in Terms of Theory of Action 703

called the "social sciences," so far as these sciences are fields of the ap-
plication of systematic analytical theory.62

As the sciences of behavior mature as sciences, they will not continue
to be the province of a plurality of competing "schools" of theoretical in-
terpretation, but will tend to converge on a logically integrated, but also
highly differentiated, conceptual scheme. The perspectives from which
the many different parts of such a scheme have been approached are
and will continue to be many, so that the process of codification can be
expected to be long and difficult. But the history of the work of codifica-
tion with which I am familiar, leads me to believe that, again and again,
what at one time have been considered competing and incompatible
schools of theory in a special field will be shown to be special cases of a
more general theory, each of which is fruitfully applicable within the
range of its own limitations. Perhaps we are not yet ready over the range
of behavior as a whole to promulgate the dictum of Schumpeter for
economics, "There are no schools of economic theory; there is only good
theory and bad theory" [31], but in my opinion this expresses the general
trend of scientific development in the behavior area as in others. In in-
terpreting this view it should be remembered as just noted, that no
theory is ever definitive but is always destined to be superseded by a
better theory; this does not usually mean that the older theory was
"wrong," it means it was limited.

Anything like the general theory of action in its present state is clearly
destined to be superseded in this sense. Furthermore, the process of cod-
ification between it and materials within its range which are not now
explicitly stated in its terms, will not merely have to proceed much fur-
ther than heretofore, but as codification proceeds, the statements both
of the theory of action itself and of the other theories will in the nature
of the case have to be modified. Granting all this as nearly obvious, I am
not aware of any drastically different conceptual scheme, i.e., one funda-
mentally incompatible, which is on a comparable level of generality,
seriously competitive in the current situation, and likely to supersede it
in a sense other than that in which, by the general process of the evolu-
tion of theory, any scheme current at a given time always comes to be
superseded.

A second point about range is very simply answered. Within any
given system type, the theory covers the whole range of microscopic-
macroscopic levels, for example, in the case of psychological systems

02 The main problem in this last connection is the status of history, which
traditionally, for the most part, has not considered the development of systematic
theory as its province though historians have, I think increasingly, begun to make
use of analytical theory developed in the neighboring disciplines and many his-
torians have made important contributions to theory.

704 TALCOTT PARSONS

from the S-R-S- system level to the total personality, in that of social
systems from the experimental small group to the total society. This
claim is fundamental to the whole status of the scheme; disproof of it
would be extremely damaging.

The next question concerns the boundaries of the general theory of
action. In terms of what are, at the same time, evolutionary "stages" and
levels of the organization of behavior I think the extremes of the range
of applicability can be defined with fair precision. At the "upper" limit,
the important consideration touches the boundary between cultural sys-
tems "culture (II)" as systems of action and two types of "nonempirical"
discipline. One such type has "existential" references but these are non-
empirical. The obvious disciplines to include here are philosophy, at
least in certain of its branches, and theology. The essential dividing line
is given by the fact that the theory of action is a theory of empirical
science. That it is not altogether independent of philosophical assump-
tions goes without saying, but this does not make it a philosophical
theory. All theories of empirical science are dependent on philosophical
assumptions, by virtue of the fact that all human knowledge is at some
level a single organon.

The other class of disciplines is central to an aspect of "culture"
which I think should be clearly distinguished from that of concern with
action, namely, what are usually called the "formal" disciplines. These
include among others, logic, mathematics, and the formal aspects of law,
Here the essential concern is with the structure of systems of meaning as
such, not their reference to the empirical phenomena of action nor to
the "motivational" aspects of the circumstances of their use. Though,
of course, linked with the theory of action in various important ways,
they do not form part of its "province."

Psychologists have more direct interest in the "lower" limit of ap-
plicability of the theory of action; this concerns the relations between
what I have called the "behavioral" organism and the rest of the con-
crete organism, summed up in the heading "vegetative organism," and
certain related questions about the physical environment and its impact
on processes of life.

As noted in the Introduction, the organic processes most closely
associated with behavior are organized in relation to the structure of
the object system of the external world; the more so the more "actively"
the organism is engaged in "coping with" its environment rather than
passively "adjusting to" it. This applies particularly to processes of distal
perception, and to locomotor processes including grasping with jaws
or limbs, etc. Proportionately to increasing dominance of such processes,
what I have called the "meaning" of stimuli and of objects of re-

Psychological Theory in Terms of Theory of Action 705

sponse becomes relatively more important than the physical and chemical
properties of the objects with which the organism comes into contact.

This analysis has required distinguishing the two levels in the organiza-
tion of the organism which I have called behavioral and "vegetative."
I have also spoken of these two levels as interpenetrating in the technical
sense of that term as used here. The theory of action must be con-
ceived as including one aspect of what is usually called the biology of
the organism, but it is not to be interpreted as including anything like
general physiology, to say nothing of biochemistry. This view comple-
ments the view that psychology cannot be a mere branch of biology
unless biology itself be conceived as the science of all life processes; in
this event, sociology, anthropology, economics, and political science must
also be treated as branches of biology.

The more general view of the relation of the evolutionary scale to
levels of selection and organization put forward in the Introduction
implies a view of the logical relations between the sciences focusing at
these various levels. Physics and chemistry, sciences which are not, in
their main subject matter, specific to life processed at all, are the
most "general" in the sense that they analyze conditions which underlie
all empirical phenomena on whatever level of organization. The higher
levels of organization do not "suspend," to say nothing of "repeal," the
laws of these extremely general sciences. The relation is rather that,
through selection and organization, special conditions of the operation
of these laws have been created which lead to "emergent" phenomena
which are not general to the whole area covered by these sciences.63

The cluster of sciences traditionally called "biological," then, seems
to deal with a somewhat less general set of phenomena, namely, those
aspects of life which are not clearly dominated by the salience of the
processes of organized, especially socially interactive, behavior. In the
sense in which physics and chemistry are general, I think it can be
said that general biology is more general than the sciences dealing
primarily with behavior. Therefore, it seems to me that a conceptual
scheme which can include both general biology and the theory of
action must be on a higher level of generality in this sense than is the
theory of action. Action then becomes a special case of the phenomena
of life, subsumed under such a more general category but having suf-
ficiently distinctive properties so that a general science of life, as a

03 Although my competence in the field is very limited, I have the impression
that, from certain points of view, physics and chemistry are now treated as two
broad types of "special case" of the same general theoretical system; their relation
is comparable to that between the psychological and sociological levels of the
theory of action.

706 TALCOTT PARSONS

theoretical system, is not adequate to such analysis, just as physics and
chemistry, as such, are not adequate to the analysis of the phenomena
of living organisms, which constitute a special case of the organization
of "matter."

The limit of applicability of the theory of action may be defined,
in the first instance, as concern with the aspects of life processes which
are not mainly associated with behavior in relation to a structured
situation of external objects. In the second instance, that limit may be
defined as concern with the aspects of the environment which do not
impinge on the organism primarily as a system of structured objects but
rather as a set of, in one sense more diffuse, physicochemical "influences."
Temperature would be a prototypical case; for physiological theory it is
not an "object" in the action sense but an "influence."64

In the light of the foregoing, the question of the limits of the psy-
chological subtheory of the theory of action does not require extended
discussion. In terms of the general hierarchy of organization and control
the upper limit stands at the boundary between psychological systems
on the one hand, social and cultural systems on the other. At the lower
limit, it is the boundary vis-a-vis the organism.

There is something like a consensus both among psychologists and
among their scientific neighbors that the central focus of the subject
matter of psychology is the behavior of the individual organism. Those
who are less analytically minded tend to include a good deal of "social
behavior" and hence raise questions of the border line vis-a-vis sociology
and anthropology. Here I would prefer to speak of "social psychology"
as, at least in part, an interstitial discipline which deals with the
boundary interchanges and the areas of interpenetration of psychological
and social systems. Historically, social psychology derives from both sides,
and I see no better reason for psychology or sociology to press an ex-
clusive claim to it any more than for chemistry or biology to claim
biochemistry. Similar considerations would apply at the other boundary,
where I would speak of physiological psychology as a discipline inter-

64 A further important problem of scope of relevance concerns the social system
level of the theory of action. Here it is necessary only to reiterate what has been
stated before, that the social system aspect of the theory of action must, on the
present premises, be interpreted to include not only "sociology" but the traditional
theoretically oriented social sciences, namely economics and political science. On
this basis it is necessary to distinguish between a general theory of social systems
and, on a lower level of generalization, the theory of specialized subsystems of
societies, such as the economy, the polity, etc. The conception of sociology has
been ambiguous in this respect. One tendency has been to define it as the general
theory of the social system, the other as the theory of a special type of social sys-
tem, that functionally concerned with integration of societies. This ambiguity un-
fortunately cannot be said to have been resolved.

Psychological Theory in Terms of Theory of Action 707

stitial between psychological and organic systems and belonging ex-
clusively neither to psychology nor to biology. In all these questions, of
course, it should be clear that the focus is on the analytical system;
empirically there is necessarily much overlap and interpenetration.

Finally, I may merely remark that I see no reason why the version
of psychological theory which can be stated in terms of the theory of
action should not cover the whole range of theoretical interests which
can be called psychological in any specific sense. This is implicit in the
statements made above about the general trend away from "schools" and
toward unification of theory. This is not to say that all important
psychological theory is at present stated in such terms. Rather, it is
a prediction about the outcome of the process of codification which is
under way. The prediction is that however present statements of the
sort attempted in this essay may be changed in the course of codification,
it is likely to produce a general scheme which, in its psychological
aspects, will be recognizable as an intellectually "legitimate" descendant
of what I have here called the theory of action.

Evidential status; prospective considerations. It seems to me that
enough has already been said about the questions involved in rubrics
18} to J^)*; further discursive treatment would be repetitious. Both the
history of the system in mediating research and the state of currently
available evidence have been touched upon, though of course in a frag-
mentary way; anything like full coverage would require a very extensive
treatise. The point which is particularly important for this system as
distinguished from others is the evidential value of the interrelations be-
tween work in the different fields to which the general theory is appli-
cable, particularly psychological and nonpsychological. Perhaps I may
illustrate with two examples of such generalization, one going each way,
which have been mentioned earlier. The first example was the paradigm
or "model" of a process of structural change in a system of action which
Olds and I first worked out for personality change in a phase of the proc-
ess of socialization and then applied (with the help of Smelser) in the
analysis of a process of change in the structure of the American econ-
omy.65 The second was the model for analysis of the boundary inter-
changes between the subsystems of a larger system of action. Smelser and
I first worked out this paradigm in connection with the inputs into the
economy from the other subsystems of the society and the corresponding
outputs from the economy. Earlier in this essay I attempted to use this
same paradigm to analyze both the inputs and outputs of psychological
systems vis-a-vis other systems of action and the inputs and outputs of

05 The model was worked out in Family Socialization and Interaction Process
[25, Chap. 4] and generalized by Bales and me in Chap. 7. The economic version
is presented in [27, Chap, 5].

708

TALGOTT PARSONS

the subsystems of a personality vis-a-vis each other. In both these cases,
benefits seem to have been gained which would not have appeared pos-
sible without the conception of a general theory applying both to psy-
chological and to nonpsychological systems. Thus, I think a very im-
portant category of evidence for the psychological version of the ap-
proach put forward here is its fit with the nonpsychological parts of the
theory of action.

Within the range of the theory of action, the answer to the question
raised concerning the value of methods, concepts, and principles of the
system outside the context of the system itself, seems obvious, if outside
the system is taken to mean outside its psychological aspect. A major
focus of the system is, of course, the fundamental unity of the theory
of action as a whole; hence the last thing which could be suggested in
the present context is the isolation of its psychological version.

Programmaticity and strategy are very closely linked. Both can
perhaps best be stated in personal terms. In this sense my own career
has had a good deal of consistency of aim from an early stage although
it has undergone a very marked development. The conception of a
theoretical system as my major focus of interest first crystallized in the
series of studies leading up to The Structure of Social Action [21]. At this
stage, it was confined to the social system case; only in the course of
the work in 1949-1950 which eventuated in Toward A General Theory
of Action [26] did the idea of a general theory embracing not only social
systems, but also psychological and cultural systems fully crystallize.

From that time on there has been a rather high level of program-
maticity in systematic procedures of codification. My program of work
has led successively into detailed explorations of the border line between
social and psychological systems, then of the status of economic theory
within the general theory of social systems, and more recently of political
theory. There has also been less thorough work (as yet unpublished),
particularly in collaboration with Dr. Clifford Geertz and others, on
some aspects of cultural symbol systems in relation to the social system,
with concentration on the symbolism of political ideologies.

My program attempts, piece by piece, to cover systematically all the
main sub areas relevant to the general theory of action. In so doing I
have tried to work in terms of three major types of reference points. The
first is the logical structure of the theoretical system itself and the
strategically important bodies of empirical data which can already be
considered to have been codified in relation to it — on a certain level
of only relative satisfactoriness, of course. The second is the establishment
of logical relations to other bodies of theoretical analysis which have
grown up independently of at least what to me has been the central
sociological core. In psychological areas, this has involved psychoanalytic

Psychological Theory in Terms of Theory of Action 709

theory above all but also, to a lesser extent, some of experimental learn-
ing theory. It has very conspicuously involved economic theory and the
far less fully codified theory available in the political field. The third
major reference is to relatively well-established bodies of empirical data
which careful analysis can place in a context of larger theoretical rele-
vance by bringing them systematically within the purview of the theory
of action.66

I have given this type of exploratory codification procedure pre-
cedence over new empirical research. As I have noted, since immense
labor is still needed in the way of selective ordering and interpretation
of knowledge we already have, there is room for legitimate specializa-
tion in such work. This does not in the least derogate the importance
of new empirical fact-finding, with the hope, of course, that this will fall
in fields of strategic theoretical importance.

Let me close within this framework. I have deliberately called this
essay an approach to psychological theory. It certainly cannot pretend
to formulate anything like a mature theoretical system. But in the psy-
chological as in the other branches of the theory of action, and with
respect to the theory as a whole, it seems to me clear that we are work-
ing in and with a genuine theoretical system, crude and incomplete as
the present stage of its development undoubtedly is. I underestimate
neither the enormous difficulties which lie ahead in the effort to im-
prove it nor the immense amount of work on the part of many people
which it will require to overcome them. But in the sciences of human
behavior, attainment of what I think the most essential of all the in-
gredients of a mature science, adequate systematic theory, is a goal close
enough to be carefully and deliberately worked for, not in a fumbling
but in a systematic way.

REFERENCES

1. Alexander, F. Psychosomatic medicine. New York: Norton, 1950.

2. Allport, F. H. Institutional behavior. Chapel Hill, N.C.: The Univer.
North Carolina Press, 1933.

3. Bridgman, P. W. The nature of thermodynamics. Cambridge, Mass.:
Harvard Univer. Press, 1941.

4. Burdick, E., and Brodbeck, A. J. (Eds.) American voting behavior.
Glencoe, 111.: Free Press, 1958.

5. Conant, J. B. Modern science and modern man. New York: Double-
day, 1953.

00 A recent example is an Interpretation of the data presented in the studies of
voting behavior by Lazarsfeld, Berelson et al. Cf. "Voting and the Equilibrium of
the American Political System," [4].

710

TALGOTT PARSONS

6. Conant, J. B. On understanding science. New Haven, Conn.: Yale
Univer. Press, 1947.

7. Cannon, W. B. The wisdom of the body. New York: Norton, 1932.

8. Freud, S. The ego and the id. London: Hogarth, 1949.

9. Freud, S. New introductory lectures on psycho-analysis. New York:
Norton, 1933.

10. Hare, A. P., Borgatta, E. F., & Bales, R. F. Small groups. New York:
Knopf, 1955.

11. Henderson, L. J. An approximate definition of fact, Univer. Calif.
Publ Phil., 1932, vol. 14, pp. 179-200.

12. Henderson, L. J. Pareto's general sociology. Cambridge, Mass.:
Harvard Univer. Press, 1935.

13. Hull, C. Principles of behavior. New York: Appleton-Century,
1943.

14. Kohler, W. The mentality of apes. New York: Harcourt, Brace,
1925.

15. Kohler, W. Gestalt psychology. New York: Liveright, 1929.

16. Kroeber, A. L. The nature of culture. Chicago: Univer. Chicago
Press, 1952.

17. Kroeber, A. L., & Kluckhohn, C. Culture: a critical review of con-
cepts and definitions. Papers of the Peabody Museum, XL VII, 1952, 47.

18. Kroeber, A. L. Knowledge of man. In L. Leary (Ed.), Symposium
on the unity of knowledge. Columbia Univer. Bicentennial Celebration (un-
published) .

19. McClelland, D. C., et al. The achievement motive. New York:
Appleton-Century, 1953.

20. Olds, J. The growth and structure of motives. Glencoe, 111.: Free
Press, 1956.

21. Parsons, T. The structure of social action. New York: McGraw-Hill,
1937.

22. Parsons, T. The social system. Glencoe, 111.: Free Press, 1951.

23. Parsons, T. Essays in sociological theory. (Rev. ed.) Glencoe, 111.:
Free Press, 1954.

24. Parsons, T., Bales, R. F., & Shils, E. A. Working papers in the theory
of action. Glencoe, 111.: Free Press, 1953.

25. Parsons, T., & Bales, R. F. Family, socialization and interaction
process. Glencoe, 111.: Free Press, 1955.

26. Parsons, T., & Shils, E. A. (Eds.) Toward a general theory of action.
Cambridge, Mass.: Harvard Univer. Press, 1951.

27. Parsons, T., & Smelser, N. J. Economy and society. Glencoe, 111.:
Free Press, 1956.

28. Parsons, T., & Smelser, N. J. A sociological model for economic
development. Explorations in Entrepreneurial History, 1956, pp. 181-204.

29. Pearl, R. The biology of population growth. New York: Knopf,
1930.

30. Piaget, J. The moral judgment of the child. New York: Harcourt,
Brace, 1932.

Psychological Theory in Terms of Theory of Action 711

31. Schumpeter, J. A. The instability of capitalism. Econ. /., September,
1928.

32. Skinner, B. F. The behavior of organisms: an experimental analysis.
New York: Appleton-Century, 1938.

33. Stouffer, S. A. Intervening opportunities: a theory relating mobility
and distance. Amer. social Rev., December^ 1940; vol. 5, pp. 845-867.

34. Tinbergen, N. A study of instinct. Oxford: Clarendon Press, 1951.

35. Tolman, E. G. Purposive behavior in animals and men. (2d ed.)
Berkeley, Calif.: Univer. Calif. Press, 1949.

36. Weber, M. Theory of social and economic organization. A. M.
Henderson and Talcott Parsons (Trans.) New York: Oxford Univer. Press,
1947.

37. Whitehead, A. N. Science and the modern world. New York: Mac-
millan, 1928.

APPENDIX: SUGGESTED DISCUSSION TOPICS

FOR CONTRIBUTORS OF SYSTEMATIC
ANALYSES1

INTRODUCTION

We will use the term "systematic formulation" as any set of
sentences formulated as a tool for ordering empirical knowledge with
respect to some specifiable domain of events, or furthering the dis-
covery of such knowledge. As is evident in science in general and
psychology in particular, such formulations may vary in their char-
acteristics over a very wide range. These variations may reflect dif-
ferences in the intentions of the systematist, limits imposed by the
nature of the subject matter, by the status of knowledge about it and
related domains, by the availability of techniques for ordering the
events in the domain, etc.

Defined in this sense, a "systematic formulation" may vary from
one or a few orienting ideas towards the conduct of research, or
towards the organization of extant knowledge within a given empirical
domain (of any scope), to an explicit, elegant, and quantified systema-
tization. Such highly diverse expressions as "viewpoint," "research
philosophy," "Weltanschauung," "exploratory hypothesis" or set of
such, "frame of reference," "dimensional system," "systematic (or
"theoretical") framework," "explanatory (or "descriptive") system,"
"hypothetico-deductive system," "theory," "explanatory mechanism"
(or set of such), "model," etc., may all be subsumed under "systematic
formulation," as we wish to use this phrase.

This study is interested in the "systematic formulations" of present-
day psychological science. Comparative analyses of "theory" and
discussions of systematic methodology have considered far too narrow
a range of formulations during the past few decades. We seek an
inventory of current systematic resources which will adequately re-
flect the diversity and richness of conceptual experimentation of recent
and present psychology. Only by the widest possible representation
of formulations, with respect both to methodological type and em-

1 This is a copy of the document concerning the discussion themes and their
significance, sent to all Study I contributors at the time of their invitation to
participate.

713

714 APPENDIX

pirical domain, can clear light be shed on problems that cut across
various classes of "system." Only in this way can problems which are
unique to given classes of "theory55 be isolated, and interrelationship
issues be treated justly and comprehensively.

This study begins with no value judgments with respect to some
preferred mode of systematization, or even with respect to some
preferred set of systematic aims or ideals. On the contrary, the only
value judgment it makes is that issues of this order have tended, in
recent decades, to be prejudged. Nor is it the intention of this study
to end with such a set of value judgments. Our intentions are ex-
plicative, not evaluative, and our belief is that explication of the cur-
rent systematic situation on a broadened and less stereotype-bound
basis is as valuable to a rational determination of next steps on the
part of systematists and research workers as it is to more effective
pedagogy.

In this era of second and rc-order self-study questionnaires and
professional nose-counting, investigators whose mode of work is as
essentially individualistic and inspirational as that of the systematist
may understandably feel that there is a suggestion of the Philistine
in any project which requires the answering of questions about their
work. To this, we can only reply that among the intentions of this
study are not eavesdropping on the creative process, the determination
of excellence by ballot, or even the charting of "directions" by con-
sensus. We believe merely that where we can go — no matter in how
many different directions — is some function of where we are, and that
the assessment of where we are can proceed perhaps a little more
efficiently in the light of the information for which this study calls.
The type of reflective re-analysis of one5s position from a common
incidence which this study seeks finds its precedent in such institution-
alized channels as symposia, anthologies, handbooks, and the oc-
casional journal issues which are devoted to a common theme.

RATIONALE OF THE DISCUSSION TOPICS

Explicit knowledge about the characteristics of the many and
varied systematic formulations put forward in the history of science
is in its infancy, but a reasonable amount of information exists about
a few of the formulations in natural science (e.g., Newtonian me-
chanics, relativity theory) particularly distinguished for their gen-
erality, explicitness, "elegance/5 and success in mediating the organiza-
tion of knowledge. It is highly unlikely that all "successful" systematic
formulations in all fields of science exhibit all of the known properties
— even in some degree — of the criterion formulations which have so

Suggested Discussion Topics 715

far been studied by methodologists. But it is probable that all formu-
lations which realize in some measure (whether actually or potentially)
such scientific objectives as "prediction/3 "understanding," or "con-
trol" exhibit at least some of these properties.

The discussion topics in the following outline have perforce been
derived from the specifiable characteristics of the class of scientific
systems which has so far received attention from methodologists of
science. Nevertheless, we have no great confidence in the adequacy
to psychology (and the biological and social sciences) of the generaliza-
tions about problems of empirical systematization made by methodol-
ogists of science. Whether systematizations of psychological data can
be expected to conform to any large number of such characteristics
is, of course, an entirely open question. Unfortunately, we do not as
yet have a vocabulary, and a set of corresponding distinctions, which
permits us to talk with precision about the widely varying character-
istics of non-natural science systematic formulations. Given writers
will therefore find that not all items will be equally relevant to their
own systematic formulations, and some items will probably be en-
tirely irrelevant. Depending on the nature of his system, the systematist
must necessarily give differential attention and emphasis to certain of
the items. He may also find it necessary to discuss the formulation
with respect to characteristics not included in the outline.

Clearly, we are aiming for commensurability of treatment, but
not blindly or rigidly so. Not only may individual writers find it
necessary to omit certain of the items, but they may wish, in some
cases, to re-interpret items in order to bring them to bear more
precisely on the nature of the formulation under analysis, and they
may wish to alter the order in which the various discussion topics
are arranged. Despite such necessary variations of treatment, the
procedure should result in a more commensurable airing of issues
connected with systematic formulations than has hitherto been the
case.

It would be meaningless to suggest any standard length for the
manuscripts. Obviously, we should like to have sufficiently sustained
consideration of the discussion topics to ensure clarity for a hetero-
geneous audience, and to derive maximum explicit benefit from the
systematist3 s wisdom with respect to the problems at issue. On the
other hand, we do not wish to burden the systematist with an overly
laborious or time-consuming task. The purposes of the study will
be adequately served by manuscripts which are as brief as is compati-
ble with meaningful discussion of the outline rubrics.

We have tried to formulate the following list of discussion topics
explicitly enough to ensure univocality of interpretation, yet at the

716 APPENDIX

same time to avoid unconscionable discursiveness in our presentation.
For reasons indicated above, we have used certain of the "standard"
distinctions and terminological counters of the general methodology
of science with a reluctance which has only given way because of the
unavailability of any alternate vocabulary for talking, with general
intelligibility, about systematic problems. If the authors of such
distinctions have, in the past, applied them in such a way as to imply
value judgments based on the degree of correspondence between the
material under analysis and the analytic distinctions at hand, we can
only regard this as a regrettable historical circumstance to which the
results of the present study may conceivably supply the proper
corrective. Indeed, a useful outcome of the present study might well
be the aid it can give towards the development of a more meaningful
way of talking about problems of psychological systematization.

THE THEMES OF DISCUSSION

11} Background factors and orienting attitudes

(a) Background factors which have influenced objectives, methods,
and content of system.

(5) Orienting attitudes which have determined systematic ob-
jectives, methods, and conceptual content.

Explanation

"Background factors" would include, of course, such matters as education,
influence of other theorists, general currents of thought within the field or
the culture at large, previous research history, or any other genetic circum-
stance which the systematist deems noteworthy.

"Orienting attitudes" register those presystematic judgments, values, and
beliefs which, in a relatively general and stable way, have determined the
aims, inductive basis, conceptual content, or formal organization of the
system. Examples might be the systematist' s general commitments towards
such issues as:

a. the nature and limits of psychological prediction

b. "level of analysis55 at which it is fruitful to constitute explanatory
constructs, with respect both to "ontological reference" (e.g., "purely
behavioral,53 "physiological," "sociological"), and "coarseness-fine-
ness" of the "causal" or explanatory units

c. utility and role of "models"

d. comprehensiveness of empirical reference (in terms of some such con-
tinuum as "unrestricted generality of scope — extreme delimitation55)
towards which it is fruitful for a system to aim, in the present phase

Suggested Discussion Topics 717

e. degree and mode of quantitative and mensurational specificity towards
which it is desirable and/or feasible to aim

/. type of formal organization (on some such continuum as "explicit,
hypothetico-deductive axiomatization — informal exposition") con-
sidered best suited to requirements for systematization, at the present
phase, in the area selected by the systematist.

In order to promote adequate understanding of the systematist's goals
and working methods, it would be desirable to make the itemization of
"orienting attitudes35 reasonably complete.

i2> Structure of the system as thus far developed

(a) Exhaustive itemization of systematic independent, intervening,
and dependent variables.

(b) Mode of definition of representative variables of each category.

(c) Major interrelations among constructs.

(d} Discussion of order of determinacy and other characteristics
of construct linkages.

Explanation

What is sought here is not a discursive summary of the system, so much as
a reconstruction of its conceptual structure via the isolation of the chief
systematic constructs of all categories, and the exhibition of how they are
interrelated within the system. The presentation need not be particularly
lengthy, since, for the purpose of the analysis, the systematist need not
summarize contents of prior expository publications, to any marked extent.

In order to promote commensurability, we are suggesting that the system-
atists adhere to the independent-intervening-dependent variable schema
which has become more or less conventional in recent methodological dis-
cussion. Since many systematic formulations have not been explicitly pat-
terned on such a schema, the recasting of the systematic structure in this
way may present difficulties, but, we suspect, not very formidable ones, in
most cases.

In cases where a systematist feels that an attempt to recast his material
into the independent-intervening-dependent variable schema does violence
to his formulation, he may, of course, recapitulate the structure of his
system in any way that he considers appropriate.

In certain cases (e.g., "positivistic" systematizations), a system may not
contain conceptual components which correspond in functional significance
to "intervening variables." In such cases, the systematist's task will obviously
reduce to the isolation of systematic independent and dependent variables,
and their interrelations.

For purposes of this study, we stipulate the following rather informal
definitions of the three classes of systematic variables.

1. The "independent variables" of a system are the terms referring to
the factors available for identification, "measurement," and, when

718

APPENDIX

possible, manipulation, which are discriminated within the system as
the antecedent conditions of the events that the system is designed to
predict.

2. The "dependent variables53 of a system are the terms designating the
classes of events that the system is designed to predict.

3. "Intervening variables" are terms interpolated between the inde-
pendent and dependent variables, having properties such that a
class of empirical relationships describable by a given number of
statements which directly relate independent and dependent variables
can be derived from a substantially smaller number of statements
which relate independent to intervening variables and these, in turn,
to dependent variables.

Note that the item -{2} discussion topics call for the isolation of "systematic"
independent and dependent variables. In explanation of this, it may be well
to note that the expressions "independent variable" and "dependent vari-
able33 have become highly ambiguous in discussions of psychological method-
ology. The independent-intervening-dependent variable schema established
(in the first instance) by Tolman for the analysis of theory implies a sense
of the expressions "independent variable55 and "dependent variable33 which
overlaps only partly with these expressions as they are used in mathematics
and in general scientific methodology. In order to be entirely clear for the
purpose of the present study, we present three senses of the expression "in-
dependent variable33 (analogous definitions may immediately be derived
for the expression "dependent variable55).

SENSE I. SYSTEMATIC INDEPENDENT VARIABLES

Terms in the construct language of a theory denoting the chief classes of
empirical events which serve as the operationally identifiable or "measur-
able,33 and, wherever possible, manipulate antecedent conditions of the
events that the theory is designed to predict. This is precisely the sense in
which the present discussion topic calls for the isolation of the "independent
variables33 of the system under analysis. We may refer to "independent
variables,33 in this sense, as "systematic independent variables.5'

SENSE II. EMPIRICAL INDEPENDENT VARIABLES

A term or expression denoting any factor in an experimental situation
which is systematically varied, or operated upon in some way, with the
intent to observe and record a correlated change in another part of the sys-
tem defined by the experiment. Sense II independent variables may be
called "empirical independent variables.33 Sense I and Sense II are very often
confused. Empirical independent variables may be specific, singular "realiza-
tions53 (operational or reductive "symptoms33) of a systematic independent
variable; they are not, however, to be identified with the systematic inde-
pendent variable to which they are ordered. Sense I independents are terms
in the construct language; Sense II independents are expressions in immediate
data language (cf. "explanation,33 item {3}). A Sense II independent vari-
able need not be a "realization" of a Sense I independent; empirical rela-

Suggested Discussion Topics 719

rions between experimental variables which are ordered to no extant theory
are often investigated.

SENSE in. "MATHEMATICAL" INDEPENDENT VARIABLES

All terms in a statement of functional dependency of which a given term
(the dependent variable) is a specified function. This corresponds roughly to
the usage of "independent variable" in mathematics. We give this rather
obvious usage for purposes of completeness.

It might be added, at this point, that in most instances systematic inde-
pendent and dependent variables are introduced into a system and given
empirical meaning by some stipulated linkage (s) to a set of empirical in-
dependent or dependent variables (this is one way of elucidating what is
meant by so-called "empirical" or "operational" definitions). Thus, in the
present analysis, a systematist may wish to employ some such distinction
when discussing such questions as "mode of definition of representative
variables" [item {2}(b)~\ and certain other questions introduced in later
sections [e.g., items {3}(c) and (d)].

i3} Initial evidential grounds for assumptions of system

(a) Identify the chief classes of experimental and/or empirical
data which have served as the initial source of evidence on which the
system was based, or have been used in any way to suggest the major
assumptions of the system.

(b) Why was this material considered "strategic, " or in some
sense "fundamental/3 relative to:

(1) other sources or varieties of data within the same empirical
area,

(2) data in other empirical areas for which the system is intended
to hold?

(c) Isolate the chief empirical independent and dependent vari-
ables (in "theoretically neutral, " "immediate data language" terms)
in the evidence on which the system is based.

(d} Show how empirical independent and dependent variables (as
expressed in "immediate data language") are linked to systematic
independent and dependent variables (construct language).

Explanation

In the discussion of (a), it would be interesting for the systematist to
consider whether, in general, the system has thus far been based primarily on
extant empirical data, or whether the systematic program has been contingent
on the prior extension, or "opening up," of a field of data by the individual
systematist, or group of investigators working within the systematic context.

In (d) we have reference to the distinction between systematic independent
and dependent variables (Sense I) and empirical independent and dependent
variables (Sense II), precisely as made above (cf. "explanation," item

720

APPENDIX

For uniform understanding of items (c) and (d), it might be useful to
specify what we mean by "immediate data language." One may say that
all empirical ("operational") definitions of a system are constructed from a
linguistic base that may be called the "data language" of the system in
question. Immediate data language is the language, presumably univocally
intelligible to all competent workers in the field, in which empirical or
operational definitions of systematic terms are put forward, and against
which primitive and derived statements of the system are compared. In
general, then, "immediate data language" tends to appear in two contexts
in connection with an empirical system:

1. in statements which are explicitly intended to provide operational
definitions of terms in the construct language, and

2. in descriptions of experimental (or general empirical) conditions,
observations, and the results of statistical or mathematical transforma-
tions of observations which the systematist or investigator is relating
in some way to the construct language of the system.

One may distinguish "immediate data language" from another sense in which
"data language" is often used in methodological discussions — i.e., as the
"epistemic reduction basis" of the terms of a system. This involves reduction
of the systematic (construct language) terms to the "ultimate" confirmation
language to which all proper statements of the system are, in principle,
reducible. We are not concerned with "data language" in this latter sense
in the present group of discussion topics.

-14} Construction of function forms

(a) How are independent-intervening-dependent variable — or, in
the case of "positivistic" systems, independent-dependent variable
function specifications constructed?

(b} Rationale of, and grounds for confidence in, the procedure.

(c) Contemplated modifications or extensions of the procedure as
the theory develops.

(d) Grounds for favoring employment or nonemployment of inter-
vening variables.

Explanation

When thrown into independent-intervening-dependent variable form,
any system will contain stipulations, at one level of explicitness or another,
with respect to the interrelations among these variables. Such construct
linkages will vary from rather general adumbrations of the functional
relationships to highly specific descriptions of function forms. Thus, "func-
tion specifications" may range from "purely qualitative" verbal descriptions
through varying degrees and modes of quantitative explicitness, depending
on the systematic intentions, the area under systematization, etc.

Such function-form specifications are, in one sense, free and creative
"constructions" on the part of the theorist. In another sense, however, they
"come from somewhere," and are "arrived at" on the basis of some set of

Suggested Discussion Topics 721

rules, however implicit. It would be most useful if systematists participating
in the present study, would make an attempt to explicate or reconstruct
their procedure in arriving at the specification of function forms. In the
case of some systems, construct interrelations may register in a relatively
direct way the interrelations among empirical variables, as determined in
specific experiments or empirical studies which are believed to have funda-
mental significance. Such relationships may be "transposed" to the syste-
matic variables in a variety of ways, ranging from empirical "curve-fitting"
to verbal descriptions of the trend of the findings. In the case of other sys-
tems, the construct linkages may apparently be arrived at by "rational
analysis," but in ways which are differentially based on inductive evidence,
and which may range in form from the positing of rational equations to the
stipulation of verbally formulated, qualitative interrelations. In still other
cases, the technique of function construction may be partly "empirical" and
partly "rational," as combined into various concrete strategies.

{S} Mensurational and quantificational procedures

(a) What procedures are either specified or presupposed by the
system with respect to the "measurement" (in the broadest sense)
of the systematic independent and dependent variables?

How would the "level" or type of mensurability presently char-
acteristic of the systematic independent and dependent variables be
located by the systematist within the terms of the logic of measurement?

(b) To what extent do the procedures for "measurement" of the
systematic independent and dependent variables satisfy the mathe-
matical requirements of whatever quantitative techniques are em-
ployed for the description of function forms?

(c) What is the systematist's estimate of the principal difficulties
in the way of increasing the mensurational and quantitative adequacy
of the system? Future plans with respect to the mensurational and
quantitative development of the system.

(d) Views of the systematist with respect to limitations, in principle,
on "level" of measurement and degree of quantitative specificity of:

(1) his own system,

(2) systematic efforts in psychological science generally.

Explanation

Obviously, certain of these discussion topics will not be relevant to
many of the systematic formulations sampled within the present study.
Some formulations will be nonquantitative, in principle. Others will be
prequantitative in their current form. In such cases, it would nevertheless
be of great interest for the systematist to discuss items (c) and (d) .

•{6J Formal organization of the system

(a) Status of the system with respect to explicitness of axiomatiza-
tion, and of derivational procedures employed.

722 APPENDIX

(J) What factors (e.g., "strategic/3 "empirical") are responsible
for the present mode of formal organization of the system?

(c) Views of the systematist about the ultimate level of formal
explicitness for which it is desirable, in principle, to aim.

Explanation

Explicitness of axiomatization and derivational specificity or rigor can
clearly vary over a very wide range, from informal exposition to detailed
hypothetico-deductive development within the resources of mathematical
notation and symbolic logic. It would be interesting if, in the discussions of
the above topics, the systematist would present his views on such questions
as the degree of "formalization" which he feels it may be fruitful to aim
towards, in areas other than those to which his own systematic work is
relevant.

In the discussion of "formal organization39 a recapitulation of the defi-
nitional techniques employed within the system would be highly useful.
Ideally, this would include a reconstruction of the roles of "implicit" (i.e.,
"postulational") definition, "explicit55 definition, empirical or "operational55
definition, and, in certain cases, "coordinating55 definition, as these are
respectively realized within the system.

{1} Scope or range of application of system

(a) Actual scope, as the system is currently constituted.

(6) Intended, ultimate scope and grounds for this delimitation.
Concrete plans and programmatic devices for extension.

(c) Interrelations, present and potential, with formulations of
other systematists in:

(1) areas coextensive with system, and

(2) other empirical areas.

i8)- History of system to date in mediating research

(a) Itemization of the chief experimental or empirical research
studies, or clusters of such, which the system has directly (i.e., by
logical implication) or indirectly (i.e., by suggestive or heuristic
guidance) instigated.

(b) What specific components of the system — e.g., orienting
attitudes, general but incompletely specified "explanatory mecha-
nisms" or constructs, specific lawful assumptions, methods — have been
responsible for the research instigated by the system?

{9> Evidence for the system

(a) Current status of the "positive" evidence for the system (to
'.the extent that this is not covered in item -{8} above).

(b) Major extant sources of incompatible or "embarrassing" data.

Suggested Discussion Topics 723,

(c) Specification of experimental designs which would be re-
garded as "critical" or important tests of principal foundation
assumptions.

(d] Types of data which, in the opinion of the theorist, the theory
accounts for more successfully than do alternate formulations. Classes
of data which alternate formulations handle more successfully.

41 0} Specific methods, concepts, or principles of the system
believed valuable outside the context of the system

(a) Methods, concepts, or principles deemed fruitful for systematic
advance in areas outside the projected range of application of the
system.

(b] Chief methods, concepts, or principles believed to be of long-
term significance, independently of the over-all structure or detailed
assumptional content of the system.

•Ill} Degree of programmaticity

(a) Evaluation of the over-all extent to which the systematic
program has been realized, at the given time.

(b) Estimation of the extent to which the system is tending towards
convergence with other coextensive systems, articulation with systems
having different empirical domains, subsumption of more limited
systems, or subsumability under more general ones.

412} Intermediate and long-range strategy for the develop-
ment of the system

(a) What classes of empirical relationships does the theory most
require knowledge about, and in what priority order?

(b) Estimate of the chief conceptual and empirical difficulties
working against the development of the system.

(c) Estimates, based on the systematist's experience, of the chief
barriers blocking general theoretical advance in psychology.

NOTE ON THE USE OF DISCUSSION
TOPIC INDEX NUMBERS

As a convenience for the reader interested in the relation of essays
to the discussion topics and in the cross-comparison of positions on
key issues, index numbers corresponding to the twelve discussion
themes have been inserted at relevant places in the Table of Contents
preceding each of the essays. These numbers are placed in brackets
immediately following the germane rubrics of the author's plan of
discussion.

By and large, correspondences between authors3 organization and
the discussion topics are straightforward, and can easily be identified
from the author's formulation of headings. Not infrequently, however,
an author's system of headings may, in one or another way, be out
of phase with the discussion rubrics, even though some or all of the
relevant issues are considered. This circumstance has led to the
following conventions:

The section designated by a given author-heading may be relevant to two
or more themes. In such cases, the brackets will contain the requisite plu-
rality of index numbers, e.g. -{3, 8, 9}.

In cases in which a section, or some part of it, is primarily relevant to a
given theme but includes brief, partial, or implicative consideration of a
number of others, that is indicated by a + after the index number of primary
relevance, e.g. «{2+>.

When a section encompasses a number of discussion topics but gives them
markedly different attention or emphasis, it has occasionally seemed worth
setting the bracketed numbers in an order which roughly reflects this, e.g.
{4, 5, 3>. Since such discriminations of relative emphasis cannot always be
clearly made, there is no implication that index numbers are not differentially
relevant when they are given in consecutive numerical order, e.g. {4, 5, 6}.

We should note, also, certain general restrictions on the use of in-
dex numbers:

With very few exceptions, they have been used only in conjunction with
major subdivisions of the papers (i.e., headings of high "value"), the excep-
tions having been mainly cases in which essays contain a final section
specifically for the purpose of bringing aspects of the preceding discussion

724

Use of Discussion Topic Index Numbers 725

to bear on the themes. In such cases, index numbers have been inserted to
identify the themes dealt with in relevant subsections.

Index numbers uniformly pertain to discussion themes as a whole, and
do not separately identify the subitems which invite differentiated discussion
under each theme. Once the correspondence with a given theme is identified,
the reader will find that, in most instances, the bearing on particular sub-
themes is easily discriminated.

In several papers, the author's plan of organization is such, as to
preclude the insertion of index numbers. In some of these (e.g.,
Ellson, Skinner) the author has preferred a type of discursive pres-
entation sans headings — or has used so few of them that any use of
index numbers would have been nondiscriminating. In a few cases
(e.g., Pirenne and Marriott, Kallmann), the author's organization
is so markedly out of phase with the discussion themes as to make any
use of the numbers either confusing or unnatural. Nevertheless, it will
be found in most of these cases that it requires little effort to determine
the author's position with respect to many of the thematic issues.
There are a few essays, however, to which certain of the suggested
themes are not relevant in principle in that the concern is primarily
with presystematic issues (e.g., Harlow).

Whatever the explicitness of relation of each paper to the themes,
it should be emphasized that each is a self-contained essay, having
sui generis properties in substance and form. Any cross-comparison
or integration of findings which the reader may wish to conduct must
depend on his own active discriminations; it will not be provided
ready-made by any mechanical device. The present system of indexing
is offered merely as a convenient starting point for comparative
analysis. It has been kept typographically inconspicuous, and used
in conjunction only with molar rubrics, both of author and thematic
organization, so as not to interfere with the organic unity of each
presentation.

Some Trends of

STUDY I. CONCEPTUAL AND SYSTEMATIC

Volumes 1, 2, and 3

EPILOGUE SIGMUND KOCH

Introduction 730

The time-reference of the study 731

Plan of discussion 732

The Intervening Variable Paradigm for Theory "Construction35 .... 733

The "strategy" for constructing intervening variable functions .... 735

Generality of intervening variable functions: achieved and in principle . . 739

The problem of "unambiguous linkage" to observables 743

Problems Concerning the Generalization Range of Psychological Laws . . 749

Revivified emphasis on problems of observation and classification . . . 750

Shift away from single species preoccupation (and related matters) . . 750

Conservatism re limits of prediction in psychology 751

Increased modesty of aim and of claimed achievement 752

The Observation Base of Psychological Science: Its Relation to the Legiti-
mate End-terms of Systematic Analysis 752

Reanalysis of S and R 755

Among S-R theorists 755

Treatment of stimulus variables in sensory psychology 759

"Stimulus" in perception psychology: the instructive case of Gibson . . 760

"Stimulus" and related variables in personality and social formulations . 761

Increased interest in perception and in central processes 764

Revivified concern with experimental analysis 766

"Presystematic" analysis of experience 767

Some transitional cases 767

"Systematic phenomenology" 768

Summary re observation base 768

Mathematization of Systematic Relationships 769

"Non-mathematical" systcmatists 771

Systematists working towards "strong" degrees of mathematization . . . 772

Formalization and Psychology 776

Belief that the hypothetico-deductive model represents scientific practice in
an incomplete and possibly misleading way; conviction that the hypo-
thetico-deductive prescription is infeasible 778

Belief that formalization is desirable in short-range future, but strong

awareness of "dangers" and difficulties 780

729

730 SIGMUND KOCH

Demonstration of some degree of achieved axiomatic explicitness in a
limited area, plus measured optimism over the prospects for extension at

comparable levels of axiomatization 780

A Concluding Perspective 783

INTRODUCTION

Thirty-six men have responded with gallantry and dedication to an
unusual challenge. In examining their own inquiring histories, they have
written history. In reflecting on and assessing their inquiry, they have
changed history. In the conjoint ordering and reordering of inquiry, they
have made history. While so doing, they have shown what analysis can
be when the creative function is not quarantined from the critical. And
in this process, the very canons of analysis which have regulated action
in our science for many years have been rid of staleness — perhaps trans-
formed.

The essays in these volumes will well repay the efforts of independent
analysts — be they interested in psychology's near history or its prospects,
in problems in the enaction of psychological science or of science. The
educator and student will find them valuable and will wish to pursue
their own analyses. Even the self-determining citizen may wish to form
his own perception of the place of psychology in the pattern of modern
knowledge. It is thus vital that nothing that could be construed as some
"official" summary or statement of conclusions stand between the reader
and the essays.

So strong has been the tendency of recent scholarship (in psychology
and elsewhere) to press complexity into stereotype and to sloganize the
subtle that the editor has been of two minds as to whether general com-
mentary should be included. Yet certain trends seem so clear as to
warrant the hope that one's impressions are not arbitrary. And when
these trends are taken together and seen against the cloth of recent
history, it is difficult not to feel that there are objective senses in which
the import of the study can lead to a profound clarification, even redef-
inition, of the ends and instrumentalities of systematic effort. It is diffi-
cult, that is, to remain silent. Yet it is meet that this epilogue be the next
thing to silence.

The present statement will thus be confined to trends which seem,
as it were, to announce themselves. Among these, it will select only a few
of the most insistent, and these few will be reported in the roughest
grain. The trends examined will not be conclusions suggested by or im-
posed on the "data" of the study. They will be more like tracings of
pervasive attitude, interest, and judgment clusters within the "data."
The interest will be in those clusters which can most clearly give a "fix"

Epilogue 731

on the position of systematic psychology relative to Its recent history.
That will entail a selection of clusters that appear most critically to
qualify or depart from the ideology and practice of recent systematic
psychology.

The Time-reference of the Study

If trends are to be seen relative to recent history, that interval re-
quires bounding. This we can do with no more sharpness than the
vaguely tapering margins of any "unit" of history. The assumption in
planning the study was a time-reference of some three decades — give or
take five years. This seems reasonable, if only for the fact that many
contributors — certainly the senior ones — are reporting on formulations
which originated, or first achieved influence, in the early 'thirties.

There is a more significant sense in which the results of the study
must be seen against the past three decades. Most will agree that during
this period there has been a relative cohesiveness in the temper of sys-
tematic psychology. Cultural history is a dangerous art, but it would be
difficult to write a cultural history of the past thirty years which did not
find them unified by a relatively homogeneous set of attitudes toward
the ends and instrumentalities of systematic work. This short epilogue is
not the place for such an effort.1

The present statement must go on the assumption that certain secrets
of recent history are open ones. Thus it is fairly evident that at some
point during the interval 1925-1930 there was occurring a transition
between an era of "schools'3 and what might be called the "Age of
Theory" — an age during which all activities were subordinate to the
production of a commodity called "theory" in a quite special sense de-
fined by the age. It will be evident that the Age of Theory was initiated
by a revivified and driving desire to ensure that psychological knowledge
become cumulative and sharable in the sense that such properties are
believed characteristic of other sciences. It will be evident that the Age
of Theory perceived the argumentation of the schools as cross-purposeful
and sterile because there seemed no agreed-upon decision procedures for
its resolution. Many other things will be evident:

The search for a "decision procedure" did not have far to go. It was
coincident with the exportation into the public domain of a bold and
positive view (or family of such) of the nature of theoretical science —
a reconstruction based mainly on certain of the outstanding achieve-
ments of physics. Logical positivism, neo-pragmatism, operationism had

1 An attempt to isolate certain of the continuities in the ideology governing sys-
tematic practice in psychology since 1930 is made in the "postscript" volume of
the series, Psychology and the Human Agent. The section of that book on the
"Age of Theory" tries to establish in some detail what must here be presupposed.

732 SIGMUND KOCH

made available a substantial body of doctrine which was open to
construal as providing a formulary for the "construction53 of theory. As
the Age gathered momentum, the belief became increasingly widespread
that the "new" generalizations of theoretical practice promised a tech-
nology for the "construction" of theory in any field. Theoretical publi-
cations in psychology tended increasingly to divide concern between
translating the new "science of science" into stipulations of the objectives
of "sound theory" for psychology, and presentation of formulations in-
tended to approximate such objectives. All will recall the development of
a dense secondary literature devoted exclusively to explanations of
proper theoretical technique, the adaptation of the new view of science
to special problem contexts of psychology, etc.

It is hardly necessary to reconstruct the atmosphere of the Age of
Theory, particularly that of its classical interval, say, from the mid-
thirties to the mid-forties. The regulation of systematic work by the
directives and imagery of hypothetico-deduction, the sub-culture sur-
rounding operational definition, the lore concerning the intervening
variable, the belief in the imminence (if not achievement) of precisely
quantitative behavioral theory of comprehensive scope, the broadly shared
judgments with respect to strategic foundation data, the belief in auto-
matic refinement and convergence of theories by the device of "differen-
tial test," the fixed vocabulary for the comparative dissection and analy-
sis of theory — all of these are easily recalled, if indeed recall is necessary.
The rather stable geography of dominating theoretical positions and the
standard contexts of apposition and opposition will also come easily to
mind. These scattered fragments define an ideology not discontinuous
with that of the present period.

At this level, caricature is inevitable. The past thirty years have seen
much change : there has been a wide and shifting dispersion of system-
atic ideas. If this were a history of the Age of Theory, it would be
necessary to document the conceptual and methodic inventions, and the
changes and cutbacks even in elements of Age of Theory ideology which
have occurred, say, in the past fifteen years. It would be necessary, for
instance, to acknowledge the shift in confidence indicated by the sub-
stitution of terms like "model" (and associated imagery) for "theory"
and, indeed, such cutbacks in aim as are registered by that from "com-
prehensive" to "limited" systematization. It would also be necessary to
acknowledge the extension of established systematic frameworks to new
ranges of phenomena, and the appearance of new formulations, new
problematic interests.

Plan of Discussion

Perhaps most perdurable and potent in Age of Theory ideology has
been a set of cognitive-strategic and epistemological agreements shared

Epilogue 733

by large numbers of investigators. The viability of these has been ensured
by their depth within the presupposition chains of inquirers. Even such
agreements represent no absolute contracts. Their construal has varied
across men and over time — but always within tight ranges. Such
"agreements" are in a sense the postulates of the Age of Theory; if not
the logical premises, the psychological ones. In this brief statement it
would be well to raise questions concerning the bearing of the study on
commitments of this order. We shall concentrate on findings which
relate to five classes of such commitments. These pretend to represent no
exhaustive classification of Age of Theory "premises;" they form, how-
ever, contexts with respect to which one can derive a first impression
of the study's general import. They are :

I. The Intervening Variable Paradigm for Theory "Construction"
II. Problems Concerning the Generalization Range of Psychological
Laws

III. The Observation Base of Psychological Science: Its Relation to the
Legitimate End-terms of Systematic Analysis

IV. Mathematization of Systematic Relationships
V. Formalization and Psychology

These commitment-classes, it must be stressed, are each no more in-
dependent of the others than can be expected of attitudes and judgments
of men.

Our task, then, will be briefly to consider each and inquire what
currents of questioning, shifts of judgment, realignments of values,
intimations of change are shaping up. The task will be bounded not only
by the limited number and generality of the trends selected but by the
mode of discussion contemplated. The aim is to exhibit trends; not to
report them in detail. Though certain trends seem to point towards some
form of "resolution35 of the problems to which they are responses, we
will not essay such interpretation at this place. Nor will it be possible to
give the position of all contributors with respect to each topic. We seek
merely to isolate major currents of change and requestioning which seem
common to large groupings of the contributors — in some cases to all of
them. If this epilogue can entice the reader back to the essays in quest
of new relationships, or prompt him to form further questions concern-
ing their bearing on history, its purpose will be well met.

I. THE INTERVENING VARIABLE PARADIGM
FOR THEORY "CONSTRUCTION"

Decisive among the commitments governing Age of Theory ideology
has been the doctrine connected with "intervening variables" and their
function in psychological theory. First introduced in the early 'thirties by

734

SIGMUND KOCH

Tolman as a modest device for illustrating how analogues to the sub-
jectivists3 "mental processes" might be objectively defined, the concept
of the intervening variable was soon after elaborated by Tolman and
others into a paradigm purporting to exhibit the arrangement of variables
which must obtain in any psychological theory seeking reasonable ex-
planatory generality and economy. As is well known, Hull in the late
'thirties identified what he had previously tended to call "logical33 or
"theoretical" constructs with the intervening variable, and in subsequent
formulations of his theory assiduously adhered to the independent-inter-
vening-dependent variable pattern.

The appeal of the intervening variable paradigm to Age of Theory
systematists was twofold. First, the criterion of "firm anchorage" of
hypothetical theoretical concepts via explicit functional relations to
"antecedent" and "consequent" observables seemed neatly to fill the
strong requirement of the age for a theoretical decision procedure. If
inferred explanatory concepts were to be unequivocally linked to ob-
servables, no longer need there be fear of irresponsible constructions
whose role within the theory is instant to the whim of the theorist (what
Hull called "anthropomorphism . . . in behavior theory" ). At the same
time, the paradigm seemed to render into orderly and intelligible terms
the problems confronting the psychological theorist: e.g., he needed
three classes of variables; he needed the interconnecting "functions"; he
needed a mode of inferring or constructing these functions; etc. More-
over, the schema was readily reconcilable with various elements of the
science of science lore which had powerfully determined Age of Theory
ideology since inception. The demand for explicit linkages with ob-
servables could be equated with operational definition. The statements
interlinking the three classes of variables could, if one so desired, be
asserted as postulates, thereby making place for the paraphernalia and
imagery of hypothetico-deductive method. The fervent drive towards the
quantification of systematic relationships characteristic of the era could
become the quest for quantitatively specified intervening variable func-
tions. And so on.

It was inevitable, then, that a lush literature develop concerning
intervening variable doctrine — a literature which, in some instances,
came close to suggesting that proper explication of the intervening
variable paradigm could provide a technology for the "construction" of
theory. Indeed, even as early developed by Tolman, and later applied
by Hull, the intervening variable schema was associated with a "strategy"
for constructing intervening variable functions. In briefest terms, this
strategy was to select or design a series of defining experiments, the
variables of which would be placed in correspondence with (that is,
"represent" or "realize") the theoretical variables whose relationships

Epilogue 735

were under determination. Standard curve-fitting techniques were to be
applied to such experimental results. The resulting equations or "curves"
were then presumably to hold for the theoretical variables whose relations
were in question. Though such a strategy can be (and has been)
elaborated in widely varying ways, its rationale has rarely been
questioned.

It cannot be our purpose even to sketch the range of issues dealt with
in the massive literature on intervening variables — or those implicit in
the actual systematic formulations which have presupposed the schema.
Here, as elsewhere, acquaintance with the general content of recent
history must be presupposed. What has seemed important to establish is
the weighty, if not central, position of intervening variable doctrine in
Age of Theory ideology. The issues raised in this context have long
ramified (and still do) into virtually every area in which questions about
systematic or presystematic procedure have been entertained. It was for
this reason that the distinctions re independent-intervening-dependent
variables were made so prominent a part of the Study I discussion out-
line. And it is for the same reason that we will be repeatedly led back
to considerations concerning the doctrine of variables in examining other
trends.

The over-all tendency of the study is to call the intervening variable
paradigm and much of the associated doctrine sharply into question,
and to do this in almost every sense in which questioning is possible.
Virtually every contributor has shown a disposition to qualify some
aspect of the doctrine: in some instances only diffidence seems to pre-
vent qualification in all aspects. Because of the scope of the issues, we
can give only the most general sense of a few findings. These we report
as they bear on three (not unrelated) questions:

1. The "strategy" for constructing intervening variable functions

2. Generality of intervening variable functions: achieved and in principle

3. The problem of Cf determinate linkage" to observables

1. The "Strategy" for Constructing Intervening Variable Functions2

The defining property of the Age of Theory — the quest for the rule
of the theoretical process — is vividly symbolized by the interest in a
strategy for constructing intervening variable functions. As we have
seen, such a strategy was associated with Tolman's original analysis, and

2 An "intervening variable function" is any functional relationship involving an
intervening variable as at least one of its terms, i.e., a relation between independ-
ent and intervening variables, or intervening and intervening variables, or inter-
vening and dependent variables.

736

SIGMUND KOCH

in fact some form of the "defining experiment" procedure has remained
a part of intervening variable doctrine until this day.

The most decisive thing that can be said about this issue is that the
originator of the doctrine has now come full-circle relative to the
feasibility of "standard" defining experiments. And, in general, his
conception of the significance of the intervening variables within his
own theory has markedly changed. It is well that this author's famous
propensity for freshening up his vocabulary from time to time not lead
the reader to take the present change lightly. For it represents something
of a bouleversement with respect to certain of the deepest attitudes
regulating the entire direction of his previous effort. Tolman's position
is best conveyed in his own words. Because of its historic significance, we
give a rather full citation.3

When it comes to the question of intervening variables, on the other
hand, this is where the schools differ. My own particular brand of inter-
vening variables were admitted to come primarily from my own phenomen-
ology. Thus Kohler's designation of me as a cryptophenomenologist was
probably correct. I do, however, attempt to objectify my intervening vari-
ables and to suggest standard defining experiments for getting empirical
pointer readings for them.

Actually, however, it should be admitted that I really have considerable
doubts not only about the practical feasibility of such experiments (since
they would involve a tremendous amount of time and labor) but also about
the validity of the results which would be obtained. (Italics mine.)

My proposal was that one should set up standard defining experiments
in each of which the obtained response or responses (i.e., performances)
could be conceived as depending primarily upon, as being a direct pointer
reading for, the variations of one particular intervening variable as this
latter is dependent upon the controlled and prescribed manipulations of one
or two independent variables. It was assumed that one could thus acquire
a sort of table showing just what the values of each intervening variable
would be as the result of such and such values of the correlated and con-
trolling independent variable or variables. And it was assumed, further,
that these relations of the values of each intervening variable to one or
two independent variables would hold in new, nonstandard, nondefining
situations as well — so that the values of the intervening variables could be
predicted from the values of the independent variables in the new situa-
tion.4 But I wish now to emphasize that this last assumption might well

8 Regrettably, we cannot make a practice of giving other lengthy citations in
this epilogue without risking its conversion into a detailed statement of findings. It
is difficult to refrain in that there are so many passages in the study which are both
historically important and quotable.

4 The defining "strategy" as given here is substantially the same as that advo-
cated in Tolman's "classical" early discussions of such issues, but it will be ob-
vious that his view as to the character of his theoretical constructs has changed.

Epilogue 737

prove invalid. For there may be all sorts of interactions between the var-
iables (independent as well as intervening) 3 in the new nonstandard situ-
ations, interactions which could not have been predicted from the results
obtained in the standard defining situations by themselves. For> in these
latter, rigid controls of all but one or one small set of independent variables
would have been imposed. Hence, I have considerable doubt concerning
not only the practical feasibility but also the validity of the proposal

It might be said, however, that there is another possible way of con-
ceiving my intervening variables. This would be to admit that they are
merely an aid to thinking ("my thinking," if you will). All anyone really
sees are the empirically stipulated independent and dependent variables.
In developing notions of what happens In between — such as beliefs, ex-
pectancies, representations, and valences and finally what I call perform-
ance vectors and their interactions — all I really am doing is setting up a
tentative logic (or psychologic) of my own, for predicting what the de-
pendent behavior should be and how it should be affected by variations in
such and such sets of independent variables (Vol. 2, pp. 147-148) .

Tolman is not without support within the present study in making
such an evaluation of "defining experiment5' strategy. Lazarsfeld has
made exactly the same evaluation in the course of his analysis of the need

In recent years (cf. "A Psychological Model," in T. Parsons and E. A. Shils (eds.),
Toward a General Theory of Action, Harvard University Press, Cambridge., Mass.,
1951), Tolrnan has moved from what was initially a "functional" conception of
his intervening variables (i.e., meaning uniquely constituted by linkages with inde-
pendent and dependent variables) towards an "hypothetical construct" interpreta-
tion. In the present article he identifies them as "hypothetical constructs" but of
a special sort; e.g., "And intervening variables . . . will have in part the proper-
ties of hypothetical constructs and not merely be intervening mathematical quanti-
ties. However, the 'surplus meanings' of my intervening variables which make
them into hypothetical constructs are not at this stage primarily neurophysiological,
as it is suggested by MacCorquodale and Meehl that they should be, but are
derived rather from intuition, common experience, a little sophomoric neurology,
and my own phenomenology" (Vol. 2, p. 98). It thus becomes appropriate for
Tolman to talk of defining experiments as "pointer readings" for intervening
variables. The early position (strictly interpreted) would have precluded such a
metaphor — though in effect the defining experiment strategy, as even then en-
visaged, presupposed that an empirical variable in an experiment could in some
sense directly mirror or "reflect" the "values" of a corresponding intervening
variable as a function of the independent empirical variable manipulated in the
experiment. The matter is, of course, academic in that it is clear that now the no-
tion of a "defining experiment" is itself a metaphor for Tolman — one which
functions as a kind of self-imposed check on his "own phenomenology" which he
claims (on excellent grounds) to "like." The defining experiment is, in other
words, a thought experiment in the classical sense of this device of the philosopher
of science. There seems little doubt that Tolman would agree that the ''defining
experiment" has always functioned for him as a thought experiment, rather than
as a workable gimmick for the manufacture of theory.

738 SIGMUND KOCH

for progress in the methodology of social science index formation. After
quoting Tolman's strategy of the defining experiment, as presented in his
1951 article, "A Psychological Model/35 Lazarsfeld points out:

The idea is that we can find one specific indicator for each intervening
variable. Everything else being constant, the variations in the indicators
correspond to the variations in the intervening variable. We have grave
doubts whether such a procedure is feasible even with animal experiments,
And we are confident that it is the wrong idea as far as the study of human
behavior is concerned (Vol. 3, pp. 482-483; italics mine) .

Later on p. 483, Lazarsfeld states:

There is just no way to develop a "standard experimental setup" or
"standard defining experiment." We will have to face the fact that to an
intervening variable there will correspond a variety of indicators and that
they will have to be reconciled in some way.

It is interesting — and indeed symbolic of the pervasiveness of major
systematic issues across the most widely disparate areas — that this critique
of one of the ruling assumptions in learning and behavior theory should
come from a sociologist.

Though most other authors do not in a comparably direct or explicit
way challenge the feasibility of the "defining experiment" procedure,
widespread convergence towards such a challenge is evidenced by many
other questions and difficulties that are raised. Thus, for instance, the
powerful attack by Brunswik (as represented by Postman and Tolman,
Vol. 1) on the assumptions of "systematic" experimental design in
general would rule out the possibility of constructing useful theoretical
relationships via the "defining experiment," as would the doubts ex-
pressed by Cattell (Vol. 3) concerning the adequacy of "univariate" ex-
perimental methods for the isolation of variables which behave as
significant "functional units" (whether these be intervening, or systematic
independent or dependent). Naturally, an observer such as Skinner (Vol.
2), whose "scientific practice is reduced to simple looking" through the
"microscope" provided by his methods, has no need for "theoretical
phantasy" nor thus for intervening variables or any strategy for their
inference. And Guthrie (Vol. 2) is not only still inclined to view his
(now greatly changed) formulation as devoid of intervening variables,
but his evaluation of many matters, ranging from the limits of psy-
chological prediction to the limited utility "of the laboratory studies of
the past generation," makes it clear that he would dismiss "defining
experiment" strategy as irrelevant to his purpose and infeasible if*

5 Tolman, E. C., "A Psychological Model," in T. Parsons and E. A. Shils (eds.),
Toward a General Theory of Action., Harvard University Press, Cambridge, Mass.,
1951,

Epilogue 739

general As we will see later, all authors in this study who work at the
levels of epistemic complexity set by the problems of social psychology
and personality raise questions concerning the status of their concepts
which suggest any "defining experiment" basis for construct inference
to be utterly beside the point.

Though on this topic as elsewhere, we restrict comment to only a
scattered sample of findings, it might be noted that both the remaining
topics in the present section have strong implications for the "defining
experiment" issue.

2. Generality of Intervening Variable Functions: Achieved and in
Principle

Whatever else an "intervening variable" may be, it is, by intention at
least, a device for -facilitating scientific statements of some generality. It
is hardly necessary to remind the reader that influential Age of Theory
formulations have often put forward intervening variable functions of
unrestricted generality, in some cases with attempts at quantitative
variable linkages.

In order to encourage concern with the generality of intervening
variable functions (and generally of lawful psychological statements) —
and especially a disentangling of intended, actual, potential, and feasible
generality "reference" — the Study I discussion topics suggested a dis-
tinction between systematic and empirical independent (and dependent)
variables. Such formulations as that of Hull's 1943 theory and Tolman's
early systems had tended to represent the antecedent and consequent
conditions discriminated by the theory as direct "observables." But the
stipulated independent and dependent variables were in fact far from
this: such notions as "past training," "maintenance schedule," "heredity"
(Tolman), or Hull's "S" or "Co" (conditions constitutive of drives)
or "G" (reinforcement) can be seen to discriminate enormously broad
and heterogeneous classes of possible "operations" and/or observations.
These are therefore systematic variables in the sense that they clearly
represent rather complex epistemic constructions made within the system
language of the theory in question. The numerous individual operations
and/or observations "designated" by each such systematic variable
would, in the terms of our distinction, be called empirical variables*

c The distinction between systematic and empirical variables, as given in the
discussion outline and paraphrased here, is regarded as a crude rendition of an
intricate epistemological picture — so intricate that fuller rendition would have
freighted the already formidable discussion outline too heavily. One must dis-
tinguish epistemic levels (or levels of abstraction) in the analysis of systematic
independent or dependent variables. A variable like the Hullian "CD," for instance,
is, if taken literally as "conditions constitutive of all drives," a complex class of
classes of classes . . . abstraction of indeterminate order. Such a "variable" would

740

SIGMUND KOCH

Now if one examines the lists of independent and dependent variables
given for such theories closely, it should be clear that the stipulated
theoretical laws are making enormously general commitments — from
the magnitude of which the theorist himself was often protected by the
tendency to equate his systematic antecedent and consequent variables
with "observables" (i.e., empirical variables). In this way it was easy to
overlook the fact that, say, an intervening variable function based on (or
verified by) values of the specific empirical variables manipulated and
recorded in a single experiment was often formulated in such a way
as to assert this function for huge and indefinite classes of empirical
variables (i.e., merely by transposing the function into the "correspond-
ing" systematic terms of the theory language) .

It would be too much to say that such optimistic or uncritical
tendencies to over-generalize systematic relationships are entirely absent
in the formulations of Study I. But there is massive evidence of a dis-
position by contributors to be far more realistic and conservative in
estimates of achieved, potential, and even intended generality of their
theoretical functions than has been the fashion in recent decades. On
the other side of this coin is a sober recognition of the fact that what-
ever degree of generality may be attainable for a given theoretical state-
ment must be purchased by progressive testing (either of a direct nature
or in terms of consequences) in situations other than that of initial
determination.

Indeed, it is of high interest that in Study I a person working within
the tradition of one of the more optimistic theories re problems of the
generalization range should make one of the most conservative assess-
ments on record of the achieved generality of intervening variable func-

denote the class of conditions constitutive of hunger, thirst, sex, pain avoidance,
etc., etc., which in turn are constructions upon the classes of alternate conditions
constitutive of each (say, hours of food deprivation or per cent weight-reduction
for the case of hunger). These latter condition-classes are in turn themselves con-
structions upon the classes of alternate procedures for the manipulation of each
(say, hours of deprivation after establishment of such-and-such a feeding rhythm,
or after satiation in such-and-such a way, or with respect to such-and-such a diet,
and so on). The level at which a theorist sets (i.e., defines or discriminates) his
systematic variables, independent or dependent, is an option of the theorist. Rela-
tive to this level, the "elements" in all lower classes in the hierarchy may be re-
garded as empirical variables (or, by ellipsis, the names or defining-property desig-
nations of such classes may be so termed). What the theorist cannot set by option
is whether the empirical variables thereby hypothesized as covarying in the way
stipulated by the theory do, in fact, c ovary. In other words, the theorist may aim
for any given level of generality, but if the aim is not to be idle, he must know
to what he is commiting himself. An analysis of the sort here hinted at helps make
such awareness explicit.

Epilogue 741

tions. In an acute analysis of the "experimental design required but
seldom used to justify intervening variables" (Vol. 2, pp. 276-280),
Neal Miller indicates that the minimum significant condition is when
there are at least two independent experimental operations and two in-
dependent measures (that is, two empirical variables on the independent
and dependent side, respectively). This indeed seems a modest demand.
Yet he then points out: "Although many behavior theorists have used
intervening variables, there are relatively few experiments which use
the design required to test and justify such variables" (p. 277).

As is generally known, Miller has in fact carried out a number of
experiments calculated to test for such modest degrees of generality. Of
these, he says: "I have been interested in applying the appropriate type
of experimental design to test whether some of the simplest situations in
which we commonly assume intervening variables can actually be
accounted for in terms of a single such variable. In many cases the
different measures show the type of agreement that would be expected
if they were all pure measures of the same intervening variable. But in
other cases, there is disagreement" (p. 278). It should be emphasized
that the situations Miller here refers to (whatever their actual empirical
complexity) are indeed simple relative to the situation-ranges for which
it has been fashionable to postulate intervening variable functions: they
typically involve a limited assortment of discrete "operations" and
"measures," respectively — a single primary drive (not "drive in gen-
eral" ) being the hypothesized intervening variable. Similarly, when dis-
cussing certain programmatic extensions of his miniature system on
conflict behavior, we find Miller saying: "The attempt to extend the
system to the types of experimental situations that will really test the
general utility of the intervening variables spotlights difficult problems
which are latent in many of our efforts to construct psychological
theories" (p. 225).

It is worth noting that in his concern with such modest problems
of generality, it is not the generality of quantitatively specified inter-
vening variable functions that Miller has in mind. On the contrary, he
takes pains to set himself off from the Hullian tradition in this regard,
pointing out that "there certainly is no virtue in the misleading trappings
of pseudo-quantification" (p. 281). And he champions "the strategy of
putting one's theoretical notions through qualitative tests first before
plunging into laborious attempts to quantify them" (p. 281). In con-
sidering extensions of the conflict theory to problems of psychotherapy
and personality, he is sharply aware of the "need for better definition
or scaling of empirical variables" even with respect to qualitative applica-
tions (p. 227). And indeed, in evaluating the applicability of the theory
to extra-animal problems, we find him saying: "In all these areas, ex-

742

SIGMUND KOCH

cept the animal experiments, rigorous testing of the application of the
theory is severely limited by the difficulty in specifying the relevant
conditions and in measuring the relevant responses, or in other words,
defining precisely the empirical data variables" (p. 231 ).

In general it is difficult to conceive of a further cry from the classical
Age of Theory attitude towards the generality of intervening variable
functions. If Miller does not propose the abandonment of defining ex-
periment strategy, he is certainly apprized of the generality-limits of
those formulations which have based what are intended to be highly
general theoretical relationships on functions deriving from the empirical
variables of single defining experiments. He may not be saying that the
defining experiment is infeasible in principle; he is certainly saying that
it has given restricted results in practice. Admirably, he has embarked
on a program of cross-checking multiple empirical independent variables
on the one hand and dependent variables on the other hand, in an at-
tempt to translate the defining experiment procedure into a device for
establishing the "qualitative" coherence of intervening variable functions
at modest generality levels.

S pence, as represented by Logan (Vol. 2), and certainly as evident
in his recent book,7 also makes a more modest estimate of the im-
mediately achievable generality of intervening variable functions than is
characteristic of classical Age of Theory doctrine. Though the attempt
to specify quantitative intervening variable functions is distinctly present,
there is a disposition to make "lawlike" commitments only as among
variables defined in terms relatively local to the data from which they
derive, and to cross-check basic relationships in a variety of experimental
contexts. Moreover, as Logan points out in conveying the flavor of the
"Hull-Spence approach" :

More abstract concepts provide greater generality but are difficult to
formulate adequately. Frequently, therefore, the intervening variables are
anchored informally at one level and more formally at another. If, at any
particular time, the theorist is not able to provide a satisfactory anchoring
for more than a limited portion of what he expects ultimately to achieve, he
may give the more general formulation as an informal suggestion to permit
trying the theoretical structure in areas outside its more formally relevant
ones (p. 310).

This can be seen as certainly a guarded position in connection with
questions of generality. As the section from which this quotation derives
makes clear, the "formal anchoring" of intervening variables to which

TSpence, K. W., Behavior Theory and Conditioning, Yale University Press,
New Haven, Conn., 1956.

Epilogue 743

Logan here refers is in terms of independent and dependent variables
defined at levels close to the data for which the function holds.

The rather conservative estimate of the achieved generality of func-
tion specifications made by the intervening variable theorists within the
Hullian and neo-Hullian tradition would, of course, be accepted and in
many cases exceeded by those (see pp. 736-739 above) who doubt the
feasibility of defining experiment strategy (whether this doubt be general
or relative to a man's own systematic problems) .

It should of course be understood that the main question at issue
here is the warranted or ascertained empirical generality of intervening
variable functions that, in fact, have been hypothesized. To conclude that
most such functions have little or no ascertained generality is not to con-
clude that their generality may not "overlap" their defining base. Indeed,
a function could prove valid over the entire universe for which it is as-
serted, even when asserted on no evidence. It is fair to say, however — and
by no means inconsistent with the trends of contributor opinion — that in
psychology the happy accidents which might eventuate in such a state of
affairs are most unlikely. Whatever generality intervening variable the-
ory, in any form coherent with the Age of Theory conception, is to
achieve will have to be won. Estes, whose systematization of learning
phenomena can be cast loosely into correspondence with the intervening
variable schema (Vol. 2, pp. 449-450), has well demonstrated how
arduous is the process of establishing even limited evidence for transitua-
tional in variance of parameters in modest cases involving determination
of parameter values for his acquisition function in one situation and pre-
dicting to a closely similar one. This he has succeeded in doing in only
a few instances (pp. 406-415).

3. The Problem of "Unambiguous Linkage" to Observables

Implicit in the present topic are questions of profound moment to
the future of psychology. Here we can develop only a few hints. Even
adequate broaching of the relevant issues would be an extensive enter-
prise.

The critique of "defining experiment" strategy is a sharp challenge
to traditional intervening variable doctrine. It deprives that doctrine of
its central recipe for the construction of theory. It implies that if the
search for a rule of theoretical construct formation is to continue, the
emphasis must shift from empirical towards imaginative rational strate-
gies: in fact it can be taken to suggest that no "recipe," no matter how
general or libertarian, is feasible. The present conservative estimate of
the achieved generality of intervening variable functions in extant the-
ories also sharply qualifies conventional doctrine. But neither challenge
is fundamental. What is fundamental to intervening variable doctrine is

744

SIGMUND KOCH

the purpose for which it was called into existence — to serve as a de-
cision procedure, a prophylactic guarantee against "irresponsible" the-
orizing. And this core component of the doctrine is clearly the demand
for "explicit," determinate, or "unequivocal" specification of all con-
struct relations; in particular the demand for "secure anchorage," un-
ambiguous linkage, as between critical intervening variables and "ob~
servables," independent or dependent.

It will immediately be seen that the demand for unambiguous link-
age to observables is nothing other than a translation into intervening
variable language of the demand for operational definition (or empirical
definition via other criteria having similar intent). Since this translation
occurs in a context in which the major problem is that of introducing
relatively high order theoretical concepts, it raises the issue of how the-
oretical concepts in psychology are introduced, validated, applied, and in
some sense made to maintain intimacy with the empirical world.

This may impress the reader as a stale and supererogatory set of ques-
tions. Do we not, after all, have an extensive backlog of secure answers
in terms of the operational criterion, the verifiability theory of meaning,
the reduction sentence, various other forms of the empiricist criterion of
meaning, etc.? If such analyses do not give us an ultimate theory of
scientific definition, can there be any question but that their general
tendency is decisive. The results of this study powerfully challenge com-
placency on such matters. Our contributors pose questions for the theory
of definition in the psychological and social sciences., neglect of which
can be rewarded by only the most fitful kind of comfort.

It is only too evident that all authors in these volumes who deal with
problems requiring constructions of considerable epistemic complexity
have persistent and severe difficulties in considering their formulations in
terms of the intervening variable paradigm. What is impressive is not
the existence of such difficulties (many will come as no surprise), but the
absolute regularity with which certain of them recur among inquirers
who often have quite different problematic concerns. Some of these dif-
ficulties are pointed up clearly and resolutely; others, though sorely press-
ing, are expressed in a more shadowy and implicative way. Some seem
in themselves of little significance; some are associated with what may
seem a "misunderstanding" of the best methodological precept. Taken as
a pattern, all are significant. Here are some items, almost at random.

a. Many men say or imply that all of their theoretical constructs are
at a "homogeneous" level with respect to any distinction that can be
made between intervening variables on the one hand and independent
and /or dependent variables on the other. They say or imply this in dif-
ferent ways. Thus, e.g., Rapaport (Vol. 3) illustrates in detail that the
same (psychoanalytic) "variables" may occur as independent, interven-

Epilogue 745

ing, or dependent, depending on the context of application. Cattell (Vol.
3) apparently takes a similar position vis-a-vis the status of variables
identified by factor analysis. Or they say it as, e.g., Newcomb, and
Rogers (Vol. 3) do, by maintaining that they do not employ intervening
variables but only independent and dependent. Or they say it, as do
Murray (Vol. 3), Cartwright (Vol. 2), Katz and Stotland, Thelen,
Parsons, Newcomb (Vol. 3), and others, by thinking of the arrangement
of their concepts on a "systems" analogy such that the array of theory-
language variables can — for purposes of the given analysis — be "entered
at any point." Or they say it by maintaining or suggesting (Rogers,
Parsons, etc.) that any of the systematic concepts can, for purposes of
the given application or research, be operationally defined. They imp!}1
or suggest it further by slurring over the distinction between systematic
and empirical variables, or specifically regarding their theoretical con-
cepts as at once systematic and empirical (e.g., Rapaport, Vol. 3, p.
110).

b. Despite ambivalence, there is a reluctance to ccuse" the interven-
ing variable paradigm. Thus Cartwright (in representing Lewin) and
Murray more or less skirt the intervening variable jargon. Katz and
Stotland, Parsons, and Rapaport seem to agree that the independent-
intervening-dependent variable distinctions give no particular insight
into the character of their theoretical formulations, but only become rele-
vant in connection with empirical applications in which sub-sets of their
concepts are linked with aspects of research situations.

c. When systematically defined independent and dependent variables
are introduced or mentioned by students of the more "complex" man-
pertinent processes, it is uniformly made obvious that these are — contrary
to Age of Theory lore — very far from direct "observables" that in fact
such variables are at an enormous distance from the scientific observa-
tion base, as conventionally conceived in psychology. Such "variables'3
are typically represented not merely as abstract, but as hypothesized the-
oretical constructions, as genotypes rather than pheno types: in short, as
something very much like intervening variables in so far as "construc-
tive," "dispositional," "inferential" status may be concerned. Indeed,
there seems little doubt that most systematists of the person and the
social context would accept without embarrassment a view of their con-
structs which held them all "intervening variables."

d. Though students of the person and the social context uniformly
and with little evident ambivalence speak the language of operational
definition, specific illustrations of definitional practice and much explicit
metatheory regarding the character of their concepts are patently and
markedly at odds with the letter of operational law. Sometimes this point
reaches the recognition threshold as, e.g., when Katz and Stotland say:

746 SIGMUND KOCH

We hold that one of the real barriers to general theoretical advance in
social psychology is the distance between genotypic constructs and our
phenotypic measures. In physics, the concept of atmospheric pressure is
fairly close to its operational measurement. In physiological psychology,
many concepts are similarly tied to their operational measurement. In per-
sonality theory and in social psychology, however, concepts like ego strength,
defense mechanisms, role systems, and role conflict are so remote from their
measurement that we have no single, clearly required set of operational
measures.

We believe this is a basic difference between the social and the natural
sciences (Vol. 3, p. 4-71) .

Examination of definitional practice — e.g., Rapaport's instructive
analysis of how psychoanalytic construct language might be ordered to a
concrete observation involving a slip of the tongue (Vol. 3, pp. 116-
121), or Rogers3 "case history" of the self construct (Vol. 3, pp. 200-
212) — will show that rules of construct application are uniformly given in
a way which leaves much to the discretion of the applier. The applier is
typically expected to discriminate the presence, absence, or "value" of
a "variable" within an intricately shifting pattern of events. The state of
affairs "designated" is presumably associated with the most extensive
range of phenotypes; moreover, the applier cannot assume that the same
phenotype is always an indicator of the same "variable" (or "value"), in
that given phenotypes are often conceived to be associated with quite
different systematic "variables" or configurations of such. The applier
must therefore simultaneously estimate and weight the "values" of a
manifold of variables on the basis of a presented observation-pattern
which (by presumption of most systematic accounts) is complex and
unique. This simple story is for the clinical case. The case for experimen-
tal application (realization) is only less "fluid" to the extent that an
arbitrary "simplifying" linkage is made which in effect puts each em-
pirical variable of the experiment at the very thin end of a vastly tapered
wedge originating at the "corresponding" systematic construct. Bear in
mind, for instance, the oft-noted circumstances concerning the purely
(and usually vaguely) illustrative character of the empirical variables
set in correspondence with systematic terms, say, in most experiments de-
signed to "validate" psychoanalytic principles. Far from validating or in
some sense sharpening the theory, it is a matter of great good fortune
when the experiment turns out — in the estimation of the proper lan-
guage community — to be illustratively relevant.

In reporting later trends, we will have occasion to note other points
which document flat discrepancies between metatheory and operational
law. Thus, here we note only in passing the unanimous emphasis of all
students of the person and the social setting — not forgetting others as

Epilogue 747

widely varied in approach as Tolman (Vol. 2), Lewin (treated by Cart-
wright, Vol. 2), and Guthrie (Vol. 2) — on so-called psychological defi-
nition of major systematic independent and dependent variables. To take
the specific instance of systematic independent variables, it is strongly
stressed by these men that specification of the principal antecedent con-
ditions of action for phyletically high order organisms involves a specifi-
cation of their inferred meaning for the organism. Concepts put forward
to meet such a requirement cannot be justly defined by "standard" oper-
ational procedures without liberalizing such procedures out of all recog-
nition— or identity.

If we may reduce to a single "trend" the most varied, searching,
tortured, oblique grapplings, the central question would seem to be this.
Granted that operational (or "reductive") symptoms for systematic var-
iables (whether independent, intervening, or dependent) form an open
class — one which can be contracted or expanded on further empirical
notice — how open a class of "observables" are we to presuppose for the
application of concepts of the type we are forced to use in saying any-
thing significant at the level, say, of human personality, social transac-
tions, etc.? Every student of the person and the social setting in this
study seems to be saying (in widely different ways) that their concepts
have essentially this in common : they are such that any application rule
which relates a given concept directly or indirectly to an observation
base of the sort specified by currently accepted criteria would involve
something tantamount to an infinite disjunction of "operations" or "re-
ductive symptoms" (or, more fully, "test condition-test result condi-
tionals"). And in fact it is more than doubtful whether even then the
meaning of the concept would be adequately conveyed.8

8 We speak here of "meaning" in so far as conveyed by empirical definition.
The meaning of a systematic concept is never "adequately conveyed" by empirical
definition alone; it is determined in a complex (and so far not well described)
way by the position of the concept in the systematic or theoretical network, the
meaning thus being constituted by other classes of definition as well: e.g., implicit
definition, explicit definition. There is much uncertainty as to the relations among
empirical and other classes of definition in scientific formulations; an uncertainty
currently manifested by the difficulties philosophers of science are having with
the explication of what is involved in the "interpretation" of formal systems. Even
more fundamental, there is high uncertainty in the philosophy of science over the
explication of each of the major definition-types, and especially empirical defi-
nition. Here it is well to note that though psychology stabilized its view of em-
pirical definition during the early 'thirties in terms of various forms of the opera-
tional criterion, and, beginning in the 'forties, in terms of a loose fusion of the
"operationist" account with such positivistic schemas as the "reduction sentence"
(Carnap, 1936-1937), technical philosophical meaning theory was in continuous
flux throughout that period, and increasingly so to the present day. The "official"

748

SIGMUND KOCH

These inquirers are either explicitly noting or implicitly responding
to the fact that their concepts and concept relations, no matter how in-
ferred or validated, "designate" complex and often subtle relational attri-
butes of observed phenomena, the "terms53 (embodiments) of which
relations are fleeting, labile, various, and easily blurred or masked by the
simultaneous presence of innumerable other "terms" of an equally fluid
character. More generally, these men are by way of asking: have we not
been premature in extending a theory of empirical definition which holds
as a useful approximation in physical science (or at one time seemed to)
to psychological and social science? In so doing, they are asking whether
it is at all sensible to expect any type of "theory" adequate to the phe-
nomena with which it deals to be subject to the kind of "prophylaxis"
presumably imparted by adherence to the intervening variable schema.

It should be noted, indeed emphasized, that the problems in the the-
ory of definition so conspicuously opened up by these inquirers are not
unique to them. Bearing in mind Tolman's doubts concerning "objec-
tive33 pointer-reading, the epistemic complexity that learning theorists
like Miller are beginning to acknowledge as characteristic of their sys-
tematic independent and dependent variables, and, indeed, the very con-
servative feelings of the same theorists re the entire question of estab-
lished generality of intervening variable functions (which means, of
course, validity "across operations3' and "across measures33 ) , it should be
clear that a problem of precisely the same order exists even at these pre-
sumably "simpler33 levels. Finally, we should note that Licklider (Vol.
1 ), in his penetrating analysis of auditory formulations, repeatedly points
to the problem of indeterminacy in the linkage between his intervening
variables and his final dependent variable as perhaps the most trouble-
some puzzle in his thinking. Even at the level of sensory theory, then,
we do not avoid such difficulties. There can be no doubt that one of the
major convergences in this study is a vast, if at the moment variably ap-
plied, pressure towards re-examination of our fundamental commitments
with respect to problems of empirical definition, and thus, of course, a

definitional epistemology in our science has thus been long out of date in its
(philosophical) area of origin. Though for some purposes it would be instructive
to explore relations between the Study I trends re definition and certain of the
newer, liberalized philosophical formulations, it would be unwise to do so in this
place. For this epilogue is committed to remaining within the data of systematic
inquiry in psychology, in the conviction that the type of methodological analysis
of primary use to working scientists must center on the work of scientists. The
final volume of the series, Psychology and the Human Agent, provides an appro-
priate context for examining the relations between philosophical and "indigenous"
methodology, and surveying certain of the pitfalls — defined both by history and by
principle — in problem solution by cross-disciplinary import, extra-disciplinary ex-
port, or indeed pursuit of the interdisciplinary common market.

Epilogue 749

revaluation of the "unambiguous linkage" criterion implicit in the inter-
vening variable schema.

II. PROBLEMS CONCERNING THE GENERALIZATION RANGE
OF PSYCHOLOGICAL LAWS

From the discussion of intervening variables, it is already evident that
concern with the conditions and limits of generality of lawful psycho-
logical statements is widespread and acute among the authors of these
volumes. This should not seem surprising. Is not, after all, the quest for
stable relationships having wide descriptive (and/or explanatory) spread
relative to the domain of study and high predictive specificity, definitive
of the scientific enterprise? Yet there is a sense in which explicit and
searching concern with the question of generality can be seen as having
momentous import. For there have been times during the Age of Theory
when the prevailing view has come close to suggesting that highly speci-
fied laws of unrestricted generality (across indicators, measures, situa-
tions, individuals, groups, species, occasions, and other conditions) could
be had almost by fiat. It is an interesting paradox that a climate in
which investigators typically reported experimental results (in scientific
journals) in the most "local" and situation-bound terms was at the same
time one in which theorists (often the same persons) translated such
findings into theoretical laws potentially adequate to "all behavior."

This lack of realism of the Age of Theory towards the conditions of
generality is manifested, of course, in the strategy of the "defining ex-
periment." It is shown in many other ways. It can be seen in the widely
distributed belief that certain limited clusters of foundation data can, if
"correctly" identified, provide the basis for "postulates" adequate to
"all" psychological phenomena. It is evident in the related belief that
such foundation clusters will result from the intensive investigation of a
limited behavior-class under conditions established by a special experi-
mental method. It is evident in the casual character of the rationale for
interspecies transposability of findings (or principles based thereon) ; in
the failure to show particular concern with checking the intraspecies
generality of findings, or indeed their transituational or any other kind
of generality, not excluding trans-experimenter generality. It is evident
in many other ways. Indeed, it is revealing that an era much preoccupied
with the analysis of "generalization gradients" as a substantive psycholog-
ical topic did not even begin to classify the "gradients" along which psy-
chological systematists generalize their findings.

If in recent years psychology has bypassed concern with a "logic" of
generalization suitable to its subject matter and phase of development,
that concern comes through as something like a leitmotiv in the present

750 SIGMUND KOCH

study. It is sounded in the most varied ways by different contributors and
in varied contexts by given contributors. Here we supplement the find-
ings and judgments re generality encountered in the discussion of inter-
vening variables by mention of a few of the additional contexts in which
relevant considerations arise.

L Revivified Emphasis on Problems of Observation and Classifica-
tion

Far from representing their formulations as advanced approxima-
tions towards comprehensive or highly general psychological theories,
many contributors in these volumes may be found showing an intense
and by no means neatly resolved interest in such pristine problems as
those surrounding the conditions, techniques, and meaningful descriptive
units for observation and classification in psychological science. To
take scattered examples: There is penetrating interest in the strategy
for achieving useful observational categories for the analysis of ani-
mal behavior under field conditions, as reported by Hinde in his dis-
cussion of ethology (Vol. 2). There is conspicuous concern with
problems of descriptive classification registered by almost all learn-
ing and behavior theorists in the study, particularly in connection with
the flurry of reanalysis (much of it of the most radical sort) of the
stimulus and response concepts — in itself one of the most instructive con-
vergences of the present study. Gibson (Vol. 1) shows a significant in-
terest in the strategy of phenomenal observation as a condition for ex-
tending the classical concerns of sensory and perceptual psychology in
a fruitful direction. Cartwright (Vol. 2) represents Lewin's constructs
(and the initial task of psychological theory generally) as in the first
instance a set of descriptive categories designed to represent psychologi-
cal phenomena in a meaningful way. Skinner (Vol. 2) advocates the
choice of "a basic datum" which will reduce scientific practice "to
simple looking." Murray, and Asch (Vol. 3) eloquently argue the need
for concepts which might prove descriptively adequate to the topography
of experience as well as that of action. Almost all contributors to Volume
3 stress the need for significant observational analysis of behavior under
"natural" conditions. Lazarsfeld represents problems of classification as
basic to the social sciences.

Virtually all the investigators mentioned in the above paragraph are
inclined to conceive of the status of psychology as still primarily at the
level of a search for significant variables.

2. Shift Away from Single Species Preoccupation (and Related
Matters)

The mere fact that these volumes make it possible to scan a wide
range of systematic interests (across both subject matter and persua-

Epilogue 751

sion), refreshingly readjusts any picture which sees the rat as monopoliz-
ing attention. But more significant is an increased disposition to assess
the limits of inter-species transposability of findings (e.g., Guthrie, Vol.
2; Murray, Vol. 3) and, within those limits, to ask realistic questions
about the strategy of such generalizations (e.g., Miller, Logan, Vol. 2).
Supplementing this is, of course, the trend towards phyletic diversifica-
tion of subjects, as represented in this study by Hinde's analysis of ethol-
ogy (Vol. 2), and by the emphasis on comparative materials of such
physiologically oriented psychologists as Hebb, and Morgan (Vol. 1),
andHarlow (Vol. 2).

In general, it can be said that only a few authors in these volumes
would disagree with Guthrie when he says: "Practically all research re-
sults in prediction, but if it is merely the prediction of how rats will be-
have under certain complicated conditions found only in a number of
psychological laboratories, we have not furthered knowledge or science"
(Vol. 2, p. 173). And indeed, only the same few would find uncon-
genial Guthrie's further statement that: "The use of the laboratory rat
almost exclusively as a subject escapes the complications of human learn-
ing, not because rat learning is essentially simpler, but because we are
protected from many aspects of it since the subjects cannot contribute
their own suggestions" (p. 193).

3. Conservatism re Limits of Prediction in Psychology

Another manifestation of concern with the problem of generality (in
this case the limits of generality) may be found in the generally conser-
vative estimates of the limits of prediction in psychology made by the
authors in these volumes. There is not only a recognition of rather severe
limits in principle imposed by well known characteristics of psychological
subject matter, but realistic recognition of the contingency of what may
be predicted on the systematic aims and methods. Moreover, there is a
tendency to become quite specific in discriminating the multiplicity of
senses in which "predictiveness" may be asserted as a property of a law-
ful or lawlike statement, and to recognize that not only the "content"
(dependent variable event-classes) to be predicted but the mode of pre-
diction is an option of the systematist. Illuminating discussion of such
matters may be found in virtually all essays, but perhaps most explicitly
in Licklider, and Postman and Tolman (Vol. 1 ) ; Tolman, Guthrie,
Cartwright, Miller, and Estes (Vol. 2) ; and Thelen (Vol. 3).

Perhaps most significant, there is a growing disposition to recognize
that certain objects of prediction may in some sense be intrinsically in-
teresting and worth investigating, while others may not. If we may again
refer to the quotable Guthrie: "A system may be productive of research,
but research has no value in itself. It is knowledge that we are after
rather than research and the test of a system is the light it throws on

752 SIGMUND KOCH

an area, and in psychology, not just the amount of prediction it makes
possible, but the ability to predict what is worth prediction" (Vol. 2, p.
173).

4. Increased Modesty of Aim and of Claimed Achievement

As one would expect from many of the points already made, it is fair
to say that the study gives general evidence of an increased modesty in
defining feasible goals of systematization, both long-range and short-
range, and in representing the extent to which such goals have been
realized. The contrast, in this regard, with characteristic Age of Theory
ideology is sharp and poignant. Claims as to global applicability of
theories or the joint achievement of high generality and "strong" quan-
titative specificity are nowhere made. Theorists having relatively broad-
scope intentions for the most part stress the narrowness of the sectors
in which these have even begun to be realized, while limited-scope sys-
tematists are showing increased interest in the relations between their
areas and others, and in moving outwards towards wider domains.

III. THE OBSERVATION BASE OF PSYCHOLOGICAL SCIENCE:
ITS RELATION TO THE LEGITIMATE END-TERMS

OF SYSTEMATIC ANALYSIS

One of the deepest sources of agreement during the Age of Theory
is in a common conception of the legitimate observation base of psy-
chological science. Here again we must presuppose rather than recon-
struct history. But it is fairly evident that, during the second decade of
this century, the "objectivist" epistemology of behaviorism achieved
strong dominance, and that by about 1930 it established virtually un-
disputed sway. Yet the transition interval leading from the era of schools
to the Age of Theory (say 1925-1930) was marked by uneasiness over
the mixed metaphysical-methodological grounds, and the inconstant
criteria developed by classical behaviorism in defense of its epistemology.
Psychology needed a clear and, so to say, "connotationally uncontamin-
ated" rationale for objectivism, one based on consistently methodological
grounds. The "operational" criterion seemed to provide this, as did later
certain of the more sophisticated formulations of the empirical criterion
of meaningfulness of the sort developed by logical positivism. The Age of
Theory has generated an extensive and varied literature attempting an
optimal rendering for psychology of one or another of the analyses of
the conditions of empirical significance made available by the newer
logic of science. And it has generated an even more varied range of
practice in response to the resulting methodological "directives."

If interpretations of technical meaning criteria imported from the

Epilogue 753

philosophy of science were free and various, certain core beliefs concern-
ing the legitimate observation base for psychological statements were
common to all of them. It is significant that these commitments were
historically prior to the importation of such meaning criteria, and that
after importation they remained untouched by the frequent and radical
changes in meaning theory, which continued in normal course of pro-
fessional epistemological scholarship.

Such rock-bottom commitments concerning the observation base
were partially characterized in considering the "determinate linkage"
problem as raised by intervening variable doctrine. Here — still in crudest
caricature — we supplement the account. Perhaps fundamental are
these:

a. All lawlike statements of psychology containing dependent
variables not expressible in, or reducible to, publicly verifiable and
thus "objectively" observable behavior indices are to be excluded as
illegitimate. More positively, this assumption stipulates that dependent
variable terms of the obsevation base designate referents which meet
the test of independent simultaneous observability by a plurality of
observers. Such dependent variable terms are to be defined in the
same observation terms as are at the basis of physical science (weak
form) , and perhaps are even translatable into or reducible to actual
descriptive and explanatory concepts of physics (strong form; more
characteristic of classical than of neo-behaviorism). The prototypical
case of an admissible dependent variable is, of course, the notion of
response, or more specifically a "measurable" index of response, in
some one of the varied, yet often unspecified, meanings of "response."

b. Similarly, it was demanded that legitimate independent vari-
ables of psychology designate referents which can pass the test of in-
dependent, simultaneous observability and are definable in either the
observation language of physical science or the concepts of physics. It
should be noted that in the case of the independent variable, the
strong- form requirement of translatability or reducibility to the con-
cepts of physics has retained more general currency (as, e.g., in the
"physical energy" criterion for the definition of the stimulus) than the
analogous requirement for the dependent variable. The prototypical
case of an admissible independent variable was, of course, the notion
of the stimulus, again in some one of many rather unseparated mean-
ings. It has often been presumed, however (especially by neo-be-
haviorists), that among admissible independent variables are also
certain intraorganismic "states" of a sort wholly or partly unspecifi-
able in stimulus terms, but such that the indicators for which, or
operations for the manipulation of which, can also be expressed in
physical observation language.

754 SIGMUND KOCH

During the Age of Theory, these assumptions were embedded in, or
rendered into, the language of the various "operational3 ' or empirical
meaning criteria imported from the methodology of science. The rather
casual character of the relation between such technical criteria and
these commitments concerning the observation base is indicated by the
widespread presumption that mere use of the language of stimulation
and behavior, "S" and "R3" entails a built-in guarantee of semantic
significance.

The commitments re the observation base of psychological science
are at so fundamental a level during the Age of Theory that no given
reconstruction can sound or be "right." Yet if anything is central to the
age, it is some such set of attitudes. A telling measure of the strength of
their hold is that individuals whose problematic interests are clearly com-
patible with a quite different epistemological rationale have often made
a point of squaring their interests with such commitments. The rather
fluid semantics over time of the word "behavior33 gives well known evi-
dence for this tendency; originally conceived as a class of events having
some functional relationship to effector processes, this term has been in-
creasingly enriched by the most various adjectival extensions (as, e.g.,
central behavior, cortical behavior, perceptual behavior, conscious be-
havior, fantasy behavior, etc.). Also significant is the fact that individ-
uals whose problematic concerns have in recent decades caused them to
bypass the dominant epistemology in important respects have given little
explicit attention to these implications of their work. In general, it is a
fact of some interest that during an interval of the science characterized
by frequent and dramatic systematic conflict, there have been few direct
challenges to the prevailing conception of the observation base.

Coherent with (if not enforced by) the behaviorist emphasis on
phenomena presumably designated by the S and R end-terms of system-
atic analysis are other characteristic aspects of the behaviorist system
of orienting attitudes. Thus, we have the well known stress on peripheral
behavior determinants generally, and the corresponding de-emphasis of
central and perceptual factors. We have the characteristic interest in
laws relating environmental stimuli to behavior and the corollary by-
passing of laws of cognition. It is difficult to say, of course, where be-
haviorist epistemology shades into behaviorist theory or pre-theory. It
is important, however, to see that we have here a relatively organic set
of commitments, one which has seemed remarkably stable over time,
being common both to classical and neo-behaviorism. These orienting
attitudes are so related that the fruitfulness, plausibility, etc., of any
given one must be some function of the fate of any and all of the others
as the behaviorist program becomes translated into action. Thus, in the
present section we place primary emphasis on findings of the study rele-

Epilogue 755

vant to the conception of the observation base, but we make limited
reference as well to factors which form a cluster with basic behaviorist
epistemology.

An outstanding trend of the study is the presence of a widely dis-
tributed and strong stress against behaviorist epistemology (both in the
narrower and broader senses just distinguished}. It is evident from all
sides. It is strikingly evident among behavior theorists who themselves
have powerfully molded behaviorist tradition. It is evident among "near-
behaviorists" like Tolman, whose methodological behaviorism has be-
come more vestigial, both in its actual effect on his theorizing and in the
role he imputes to it in his metatheorizing. Many of those whose prob-
lems have not been set by the emphases of behaviorism seem more dis-
posed than formerly to question the adequacy of its epistemology — even
if they are not always ready to relinquish the "objectivist clang" of the
independent and dependent variable language which has seemed so
necessary a condition of respectability in recent decades.

In sampling a few of the trends, we begin with ( 1 ) the radical re-
analysis of stimulus and response ( and related developments concerning
the specification of "basic" systematic independent and dependent vari-
ables) evident in many of the essays. We then briefly consider (2) the
generally increased interest in perception and central process on the part
of certain S-R theorists, and (3) the evidence for a revivified concern
with experiential analysis.

1. Reanalysis of S aad R

a. Among S-R theorists. Though stimulus and response have in
some sense always been under reanalysis, it is rare that this enterprise has
proceeded with the abandon evident in the present study. The radical
flavor of the trends towards reanalysis of the stimulus and response con-
cept is well conveyed by these words of Neal Miller :

In general, stimulus-response psychologists have tended to bypass prob-
lems of the type we have just been raising [the definition of S and R]. By
intuition and trial and error, they have concentrated on experimental situa-
tions in which the stimulus and response were so simple and manageable
that the lack of more precise definitions or laws concerning these variables
was not a practical problem. Using such situations, stimulus-response psy-
chologists have concentrated on determining the laws governing the con-
nections of responses to stimuli. Thus, stimulus-response psychologists may
be said to know and care relatively little about either stimuli or responses;
they are specialists on the hyphen between the S and R and could more
aptly be called fehyphen psychologists," or to use Thorndike's term, "con-
nectionists" (Vol. 2, p. 242; italics mine).

756 SIGMUND KOCH

The direction in which current reanalysis is tending is well symbol-
ized in some incisive and brilliant paragraphs of Guthrie's essay. Since
Guthrie is there evaluating assumptions concerning the treatment of S
and R of an entire generation — one which includes himself as a dis-
tinguished member — it is well that we quote him rather fully:

What, then, will be the terms in which general rules of behavior are
stated? What will constitute the antecedents and what the consequents?
What can we look for as stimuli and what as responses? . . .

In the physical sciences,, this problem has been solved by minimizing the
role of the observer and by the use of conventional instruments of measure-
ment with the attainment of a high degree of objectivity, . . .

In psychological observation it would,, of course, be a great advantage
to reduce both stimuli and responses to this point of high agreement. But
there is reason to believe that this reduction cannot ordinarily be obtained
in psychology. The phenomena in which the psychologist is interested are
not specified in terms of mass, length, and time. They involve categories
not reducible to position on a scale. In fact, they normally involve patterns
of situation and movement that require recognition by a human observer,
and this recognition is of an order indefinitely more complex than the
recognition of relative position involved in comparing a length with a
scale (Vol. 2, p. 164; italics mine) .

During the first century of the development of psychology, we have
made great efforts to be objective. We hoped to achieve this by limiting
ourselves to the categories of physics and using as the weather signs of be-
havior only the physical or chemical events normally activating sense
organs. The determinations of absolute thresholds in the various senses, the
hope that response could be treated just as movement in space which was
the crude interpretation of behaviorism, failed to carry us very far toward
the understanding of behavior. The reason for this is that we cannot reduce
the classes of psychological facts which make up the data we must deal with
to component movements in space. Patterns of stimuli and patterns of re-
sponse have their psychological significance and usefulness tied to their
patterning — pattern as pattern must be recognized and dealt with. Ma-
chines can be devised to respond to pattern, but the human observer re-
mains the only practical tool we have for the recognition of patterns in their
variety and multiplicity, . . ,

The history of our effort to use as the weather signs of behavior simple
physical or chemical changes involving sense organs is an interesting one.
One difficulty it encountered was that these stimuli did not always stimu-
late . . . But the real failure goes deeper. The patterns of physical change
that occasion response, we find ourselves inevitably describing in perceptual
terms. It is not enough that they be available in the physical situation nor
is it enough that the organism's attention orient sense organs to receive
them; it is further necessary that they have meaning for the responding
organism (p. 165; italics mine).

To object to treating a simple physical change as a stimulus or a

Epilogue 757

muscular contraction as a response is not to deny that all stimuli are ana-
lyzable into such physical changes or to deny that any specific response is
analyzable into muscular contractions and glandular secretions. That should
be assumed. It is, however, a denial that the psychological description of
behavior can be made in physical terms. It requires psychological terms
which will name recurring patterns of physical change usually requiring
identification by an observer which will include recognition of their stimulus
value usually judged by time relation to the response (p. 166; italics mine).

It is already evident that the leading proponent of contiguity theory
and a leading representative of reinforcement theory agree in making a
severe judgment of the practice of past decades in explicating the nature
of "stimulus35 and "response.35 Beyond this, there is much in the positive
analyses of Guthrie and Miller that is consonant.

The "liberalization" of S-R theory put forward by Miller in his essay
is very liberal indeed. Nowhere is this more clear than in the discussion
of his method of "functional behavioral definition.3' In this sequence
(Vol. 2, pp. 238-242), Miller draws together and states more boldly
than ever before certain tendencies in the treatment of S and R long
present in his work. In his present statement, "a response is any activity
by or within the individual which can become functionally connected
with an antecedent event through learning; a stimulus is any event to
which a response can be so connected" (p. 239). It is clear from the
discussion at this place that, like Guthrie, Miller is disposed to cut S-R
theory off from any "physical energy3' criterion of the stimulus. It is
clear, too, that Miller, like Guthrie, would consider the determination
of any given S or R as an experimental or empirical-observational prob-
lem. It is clear, also, that both would now think of S or R in an abstract
and, so to say, "ontologically neutral" way as corresponding to any ante-
cedent condition which can be shown to have a stable relation to any
consequent, or vice versa. There would be further accord in acknowledg-
ing that such antecedents and consequents are variable event classes
(different instances of which occur from occasion to occasion), and that
they are in every case constructions or discriminations made by the ob-
server.

What is not equally clear (though by no means ruled out) is
whether Miller would put the same stress as Guthrie on the need to
specify stimuli in "perceptual terms" — in terms which acknowledge that
"it is ... necessary that they [stimuli] have meaning for the respond-
ing organism" (p. 165; italics mine). On Miller's conception, this might
or might not emerge as a constitutive property of stimulation from ap-
plication of the method of "functional definition."

Though in other S-R formulations of the study we do not find evi-
dence of comparably radical departures from previous practice in the

758 SIGMUND KOCH

treatment of S and R5 there is still definite responsiveness to certain of
the difficulties bequeathed by earlier treatments. Thus Logan, in his
reconstruction of the empirical independent variables for the Hull-
Spence approach, cites a variety of "physical energy" examples as con-
stitutive of the systematic independent variable, "stimulus" (Vol. 2, p.
315). But after detailing Spence's distinction between the "situational
stimulus" (which "can be described in such physical terms as frequency,
amplitude, wavelength, etc.") and the " 'proximal3 or 'effective' stimu-
lus" ("that fraction" of the situational stimulus "which is perceived at
any one time" ) , he adds :

The effective and situational stimuli are not isomorphic, but the rules
by which one determines the effective stimuli from knowledge of the situa-
tional stimuli are not yet very fully understood (p. 314) .

With respect to response, Logan quite illuminatingly points out:

As with the independent variables, adequate definitions of the system-
atic dependent variables have not been made at a high level of abstract-
ness. There are formidable problems to be resolved when one attempts to
formulate such definitions. For example, latency is the time required to in-
itiate a response. If one is running rats in an alley, one must decide how
rigidly to control the subject's orientation at the time the door is opened,
whether to provide any distinctive ready signal, and when to define the
response as having begun. Certainly one's measures are affected by these
decisions. And comparable problems of selecting criteria are involved for
each response measure (p. 326; italics mine) .

Perhaps the most conservative note struck by any S-R theorist in the
study is Estes3 statement that "... by stimulus and all variants of the
term I refer to environmental conditions describable in physical terms
without reference to the behavior of the organism" (Vol. 2, p. 455).
This note is at once modulated in the next sentence which defines
"stimulating situation" as "all sources of stimulation that are mentioned
in the experimenter's description of the experimental situation . . ."
And it is further modulated by the fact that in Estes3 theory the "S" is a
set-theoretical concept conceived as a statistical population of elements
which fluctuate from trial to trial. It is this (hypothetical) population
of elements which is conceived in "physical terms without reference to
the behavior of the organism." In an application of the theory, the set-
theoretical construct is placed in correspondence with an experimenter-
described "stimulus situation," the accuracy of which "we must, ini-
tially at least, assume3' (p. 456). In the Estes formulation, the experi-
menter-discriminated stimulus situation is the empirical independent
variable-class constitutive of the "physical33 S as a systematic variable.

Epilogue 759

To this run-down of S-R theorists' positions on S and R, a final
word should be added concerning Skinner. Skinner has not in his essay
(Vol. 2) addressed the present issue. But it should certainly be acknowl-
edged that Skinner has an ancient priority on the view that dissociates
S and R from physical energy or specific movement criteria, and sees
them as both experimenter-discriminated and experiment-defined. How-
ever, other trends of the present analyses such as Guthrie's emphasis on
meaning, on perceptual specification, and in general on the importance
of qualitative observation by human agents (in contradistinction to
automatic recording), and Miller's utter flexibility in conceiving S and
R to be applicable to all antecedents and consequents whatsoever (in-
cluding central process and other hidden matters) — such trends had not
been anticipated by Skinner. Nor is it likely that he would condone
them.

We have considered the ferment of reanalysis, among S-R theorists,
of S and R of such high interest as to merit extended treatment. But its
true significance can be seen only within the entire pattern of findings
germane to S and R. Here we can only briefly give the drift of wide-
spread analysis by many authors. To simplify the task, we restrict con-
sideration to findings concerning the "stimulus."

b. Treatment of stimulus variables in sensory psychology. The sen-
sory psychologists in the study (Licklider, Graham, Pirenne and Marriott;
Vol. 1 ) have a message of profound importance for behavior theorists
and indeed all concerned with the systematization of "post-sensory"
processes. As cogently stated by Conrad Mueller (Vol. 3, pp. 791-797),
it is that in sensory psychology there is great diversity in mode of specifi-
cation of stimulus variables — diversity in specificity, in complexity (i.e.,
"the length of the chain of definitions linking them to experimental pro-
cedure" ) , and in other ways.

Mueller emphasizes that "perhaps the greatest diversity" is in "the
extent to which the experimenter adopts the language of physics," in-
dicating that stimulus language in technical experimental contexts can
range from the "extra-scientific language" of ordinary "objects" to the
language of physical dimensions, and indeed to extensions of these
dimensions (including "more and more of a commitment to theory,
either physical, physiological or behavioral") prompted by the require-
ments of specific problems. He also stresses that stimulus specifications
can "differ in terms of the extent to which the organism is involved in
the definition," some stimulus terms being describable without such
reference, but many others (e.g., the specifications of color stimuli) re-
quiring incorporation of complex combinations of behavioral and
physiological data and guesswork.

760

SIGMUND KOCH

The general picture that emerges is one of thoroughgoing contextual-
ism in mode of stimulus specification — a contextualism which always
represents a creative response to the requirements of specific problems.
The sensory investigator deals with his major "antecedent conditions"
not at one homogeneous, prefabricated physical level, as current stereo-
type concerning sensory psychology would have it, but at varied and
shifting levels which are always problem determined. The airing of this
message, so copiously documented by the sensory contributions, in mixed
company could have a most liberating effect on systematists in other
areas. It could embolden them to be similarly contextual and problem-
centered in the identification and analysis of their independent variables
(empirical and systematic). It suggests further that the definition of the
"stimulus" (i.e., as a systematic variable), or the "constitution of the
S-class" is not, as ordinarily conceived, one problem but many problems
— that even within a given systematic formulation (particularly if it lay
claim to reasonable generality) it may be idle to seek some single de-
fining property or specification-mode for S, or indeed any other "class"
of antecedent conditions playing an analogous role within the formula-
tion. If this message is liberating, be it noted also that sensory psychology
stands as an impressive reminder that conceptual freedom and discipline
are not incompatibles.

Sensory psychology contains other lessons of importance for the out-
side. In reading an essay like Graham's on color vision (or, for that mat-
ter, Pirenne and Marriott on the quantum analysis of brightness vision,
or Licklider on audition) it is only too obvious that sensory psychology
has won its way to appropriate degrees of specificity in the identification
of independent variables slowly and painfully and via the most complex
interactions — sometimes cumulative and sometimes discontinuous — be-
tween hypothesis formation and experiment. It is also obvious that de-
spite enviable advances in this oldest branch of psychological science, a
vast range of problems, including ones of stimulus specification, is vastly
open. We do not pretend that difficulties of the same type will confront
systematists in areas the analysis of which demands independent vari-
ables of quite different types (including ones to which the metaphor of
"stimulus" cannot be justly stretched) from those requisite for the an-
alysis of sensory mechanisms. But the experience of sensory psychology
should serve as a sobering reminder to those who anticipate sweeping or
rapid progress in systematic formulations which aim towards even
moderately ambitious combinations of generality and specificity.

c. "Stimulus" in perception psychology: the instructive case of
Gibson. If problems of stimulus specification as posed by classical sen-
sory psychology are still highly open, the perception psychologist, Gibson
(Vol. 1), reminds us that there may be an indefinitely large range of

Epilogue 761

questions concerning stimulus specification of a type not only open, but
unposed. Gibson points to the possibility that there may be many di-
mensions of stimulus specification — the "higher order variables" of
stimulation — which have not been touched by classical sensory psy-
chology. And he makes this possibility concrete and convincing by
identifying certain variables "of adjacent and successive order" as, e.g.,
gradients specifying the "texture35 of the "optical array," which seem re-
lated in orderly ways to variables of experience and reporting behavior,
some of which latter have also eluded discovery by classical psycho-
physics. Others of these dependent variables are of a sort which, if pre-
viously recognized, have not traditionally been regarded as in any direct
way controlled by variables of stimulation. Gibson thus depicts for us
the outlines of a heretofore neglected science of stimulation — one which
could have profound consequences for a rephrasing of traditional ques-
tions of psychology.

Gibson's program, of course, is that of achieving, by combined
strategy of phenomenological analysis and experimentation, a specifica-
tion of dimensions of perceptual experience and behavior in physical
stimulus terms. But it should be stressed that in Gibson's view, much of
the relevant physics does not exist: "... the physics appropriate for
the study of the perception of surfaces remains undeveloped" (p. 470).
Nor, by the same token, does a suitable metric exist for many of the
"higher order" stimulus variables. Yet the "simple co-ordering of judg-
ments to stimulus variation can proceed without the sophisticated pro-
cedures of modern psychophysics" (p. 499). We thus have a bold pro-
gram which looks towards a vast extension of the range of physically
specifiable "stimulus" variables, but one which by seeming paradox (as
against behaviorist epistemology ) can only be advanced via experiential
analysis.

There are several important morals here for problems connected
with the legitimate observation base. Perhaps most instructive is the fact
that Gibson, mainly on phenomenological grounds, looks forward to a
far-reaching extension of the range of independent variable conditions
which could come under physical specification, at precisely the same
time that behavior theorists, on "behavioral" grounds, have become
definitive in their relinquishment of a physical energy criterion of S. Any
residual tendency of people to see a fixed tie-up between behaviorist
epistemology and physicalism on the one hand, and phenomenologism
and some heady type of anti-physicalism on the other (a view not in any
case justified by history) can find little nourishment in this circumstance.

d. "Stimulus" and related variables in personality and social
formulations. Formulations of the person and the social setting have
been somewhat more free of a "stimulus" idiom for the characterization

762 SIGMUND KOCH

of major independent variables than those less directly addressed to
"complex" man-pertinent processes.9 Nor is it easy to generalize practice
in these areas, because of the great variability of systematic aims and of
conceptual posit. The fact that recent methodology has tended to bypass
systematic experience in these areas is regrettable in that it could well
be that preoccupation with man-relevant problems has led to certain
lessons concerning problems of identifying and phrasing important inde-
pendent variables which could not otherwise be learned. It is thus un-
fortunate that fuller consideration of the implications for this issue (and
related ones) of practice in the epistemically more "complex" areas must
be reserved for another occasion.10

It should first be noted, as part of the picture of variability, that in
certain of the formulations there is relatively little concern with inde-
pendent variables of a sort functionally analogous to "stimulus" notions
(e.g., environmental inputs). This lack of concern is less a matter of
principle than it is one of problematic priorities and perhaps practical
feasibilities. Thus in formulations like those of psychoanalysis (Rapaport,
Vol. 3), Lewinian personality theory (Cartwright, Vol. 2), client-
centered therapy (Rogers, Vol. 3), and perhaps the Katz-Stotland type
of attitude formulation (Vol. 3), the interest at this phase is mainly in
the working out of concepts — often of rough grain — for the specification
of intra-personal processes. In some of these theories, notably the Rapa-
port version of psychoanalysis and the Cartwright version of Lewin,
there is keen appreciation of the neglect of relations between the intra-
personal processes and structures on the one hand, and the "objective
environment" on the other. In the case of other formulations represented
in the study (Asch, and especially Murray, Vol. 3) there has been much
explicit interest all along in problems having to do with the conceptu-
alization of environmental relationships.

The major generalization that can be made is the striking conver-
gence— if not coalescence — of the current "liberated" S-R handling of
the stimulus, and the common-denominator of practice among person-
ality and social systematists in the treatment of environment-variables.
To those who have seen the methodological problems of psychology pri-
marily in terms deriving from the study of learning theory, it will be
illuminating to discover that the present "convergence" is largely uni-

9 This problem-instigated departure from convention has not in general been
paralleled on the dependent variable side, where the language of "behavior" in
some one of the many Inflated forms of its initial sense (i.e., as being some func-
tion of effector activity) has wide currency.

10 Such matters are more extensively addressed in Psychology and the Human
Agent (Vol. 7).

Epilogue 763

lateral: it Is the S-R theorists who have moved and the man-preoccupied
systematists who have (relatively) stood stiU.

Thus, for instance, it is most instructive to read, with the spirit of
the present Guthrie-analysis of S in mind, the brief sequence (Vol. 3,
pp. 26ff.) in which Murray recapitulates his concept of "press35 and
"latent-press." After persuasively developing the point that "it is not
so much the physical attributes as such but the known or supposed man-
pertinent capacities of objects which influence behavior," Murray in-
troduces his press notion as the basic environment-specifying concept in
his thinking. The press is a "subject-pertinent" property of an object
(animate or inanimate, but most typically at Murray's level of analysis
a property of an "alter"). It is, of course, jointly constituted by an en-
vironmental object-property and the perceptual or, as Murray would
say, "apperceptual" processing of the subject. Translated into a "stimu-
lus" idiom, the press could be characterized as a stimulus-cum-meaning
relative to a specific subject. The "objective" definition of a press would
be via an apperceived construction by "the psychologist, by selected
judges, or by the conventional majority" (p. 27). This can, of course, be
distinguished from the subject's protocol or the subject-definition as in-
ferred by less direct means (e.g., projective tests, behavior indices and
contexts, etc.). Comparison of such a treatment of "environmental"
antecedent conditions with the extended quotation already given from
Guthrie will show agreement on virtually every point. Nor is there any-
thing flatly incompatible between Murray's account of "press" and
the tendency of Miller's present analysis of S.

The same types of requirements which Murray attempts to meet in
the concept of press (and related considerations) are responded to in a
variety of ways and at varying levels of explicitness by all other students
of the person and the social setting. The important fact is that they are
universally acknowledged. This is manifested, for instance, by the strong
emphasis of the systematists in this group on the need for "psychologi-
cal" definitions of "environmental" variables at levels of analysis suit-
able to their problems. There is general recognition that "environ-
mental" variables must be specified as systematically complex entities,
most of which (for purposes of the problems engaged by these men) are
artifacts of a human environment which embody either conventionalized
or idiosyncratic meanings and which, moreover, enter the "causal equa-
tions" of experience or action as mediated by specific perceptual and
cognitive processing by the "responding" organism. It seems agreed also
that the "values" of such variables can be assigned in only one way : by
individual or pooled experimenter (or observer) inferences which are
themselves "perceptual" discriminations of a most complex sort.

764

SIGMUND KOCH

It is clear from the contributions in these areas that little progress
has been made with regard to a "logic" of environmental variable-
specifications. Whether there is a "logic" or a set of such, other than
rough and ready rule-of-thumb solutions to such matters, would seem
a genuine question. Certainly no general definitional device or paradigm
will handle the relevant problems. Indeed, we have seen that a general
solution is precluded even at the level of sensory psychology, one of the
few areas in which the use of "stimulus" language does not involve an
abuse of metaphor. The contextualism of sensory psychology in its modes
of specifying independent variables must be multiplied many times over
at the level of analysis occupied by current "behavior" and learning
theory. And this value must be multiplied by a still more generous factor
at levels corresponding to the manifold interests of personality and social
theory.

2. Increased Interest in Perception and in Central Processes

Here we restrict attention to certain dramatic changes, as against
earlier Age of Theory practice, in the extent and character of the con-
cern of S-R theory with perceptual, cognitive, and central processes. As
is evidenced by recent widespread interest in problems of perceptual
learning, such changes have been under way for a number of years. But
they are documented with particular force in this study — especially in the
papers of Guthrie and Miller.

It is well known that earlier neo-behaviorists tended to bypass con-
cern with perception and other "central" phenomena at the level of
primary principles in the hope that such matters might ultimately be
dealt with as secondary or "derived" phenomena. A transition seems to
be under way towards regarding these problems as requiring analysis at
the very foundations of the scientific enterprise. For Guthrie, such an-
alysis not only is taken as necessary to the proper conceptualization of
the end-terms of systematic analysis (S and R), but also is seen as in-
tegral to the specification of any fundamental law which can aid in
identifying the conditions under which learning takes place. Though
Miller's approach does not demand that the latter condition be met, it
certainly suggests that any "definition" of S and R must be so set up as
to provide for the expectation that S will often be such as to require
perceptual specification and to acknowledge the possibility that final R
will often be contingent on complex central or cognitive processing.

To turn first to the Guthrie position, the seriousness of his redefini-
tion of S and R is vividly registered in the radical alteration of his major
(some would say his only) "rule" for identifying the conditions under
which learning takes place — the principle of association (Vol. 2, pp.
185-189). The character of the change is evident from the short form

Epilogue 765

of Guthrie's new principle which reads "what is being noticed becomes
a signal for what is being done" (p. 186). The corollary emphasis of
Guthrie on the analysis of "attention" ("Attention becomes, in the
present account, the point at which learning occurs") certainly betokens
a profound shift in the direction of acknowledging central determinants
of behavior, even if Guthrie prefers to hold on to the peripheralist
language of stimulus and response for the phrasing of relationships at
such "central" levels. Further, if indirect, acknowledgment of the in-
fluence of intraorganismic processes (not excluding, one takes it, central
events) is revealed in Guthrie's statement that:

The complexity of the nexus of determiners of action requires that pre-
diction allow for high degrees of error. Most instances of associative learn-
ing are cited after the fact, and do not constitute prediction (p. 189).

Turning now to the Miller position, it is well first to indicate that he
is still soberly disposed to stay within the neo-behaviorist framework of
orienting attitudes re central process. Thus, for instance, he says: "Al-
though recognizing that much of man's behavior involves cognitions, I
have preferred the strategy of trying to explain such behavior as the
outgrowth of simpler, noncognitive mechanisms" (Vol. 2, p. 262).
Miller's interests in the phenomena of cognition, as shown in the present
mobilization of his thinking, are in fact so extensive as to make the
preceding disclaimer distinctly necessary. The flavor of his approach to
such matters is given in his discussion of "Thinking: central cue- and
drive-producing responses" (pp. 242-248) . He there points out:

One of the most important advantages of functional definitions of stimu-
lus and response is that such definitions can be applied to central as well as
to peripheral events. Instead of emphasizing anatomical location, our defi-
nitions direct attention toward the more significant problem of functional
laws. These definitions free the S-R theory of thinking from being restricted
to proprioception, allowing the theory to exploit images., response-produced
drives and rewards, perceptual responses, perceptual learning of acquired
distinctiveness or similarity, and the possibility that central responses can
contribute to the focusing of attention (p. 242; italics mine) .

The character and range of the interest in central processes is further
illuminated by Miller's interesting discussion of "Relational responses to
relational cues" (pp. 248-252) which he acknowledges as having
importance in much behavior. In a brief section concerning "Multiple
cue and response potentialities," Miller gives further evidence of the
extraordinary flexibility of his conception of S and R :

By now it should be apparent that I believe that most stimulus objects
present the organism with a multiplicity of potential cues. Thus, the organ-

766 SIGMUND KOCH

ism may learn to respond to the absolute position of a stimulus object, to its
relative position, to its absolute brightness, to its relative brightness, to its
color, to its form, to its being the one object that is different, etc.

In the following paragraph he says:

Some of these cues may be relatively direct products of end-organ stimu-
lation, others may be the result of various levels of innate mechanisms for
analyzing and processing such stimulation, and yet others may be the result
of learned cue-producing responses (p. 251 ; italics mine) .

From all the above, it should be clear that Miller is not only justified
but documenting an important historic trend when he says (p. 243):
"It is obvious that the postulation of central responses, such as per-
ception and imagery, reduces the gap between S-R and cognitive theory.55
When, in the next sentence, he says "there is still a difference in that
we clearly assume that these central processes follow the same laws as do
peripheral stimuli and responses," one can only wonder to what extent
nature will cooperate with man's verbal preferences.

3. Revivified Concern with Experiential Analysis

Whether behaviorist epistemology is logically incompatible with the
systematic utilization of experiential analysis is a question much in need
of clarification, but one beyond the scope of this epilogue. It may
indeed be possible, on some interpretations of the role of verbal report,
to make a plausible case for the thesis that all meaningful questions
having a presumptive experiential frame of reference can be dealt with,
in principle, by behavioristic methods. But no one can deny that for
more than forty years behaviorist epistemology has had the pragmatic
effect of fostering a set of attitudes which tend to either devalue or
divert attention from most problems which, by virtue of historical or
extra-scientific associations, have an "experiential odor" — quite inde-
pendently of whether the investigator believes the problem compatible,
in principle, with behavioristic methods.

An important and quite general trend of the essays is an increased
recognition of the role of direct experiential analysis in psychological
science. This trend cannot be represented as a kind of phenomenologistic
revolution. It is expressed quite variably, usually in tones which are
either cautious, oblique, or qualified, and most often in a way which
grants experiential analysis only the second-grade legitimacy of an
accessory device to hypothesis formation. And even when experiential
variables are considered legitimate or indispensable elements of system-
language, there is relatively little positive consideration of the definitional
and other methodologico-strategic questions thereby introduced. But the
trends are definite and could be prognostic.

Epilogue 767

The general pattern of findings can best be indicated against the
distinction implicit in the last paragraph between presystematic, "acces-
sory" uses of experiential analysis and explicitly systematic uses for pur-
poses of constructing variables having undisguised experiential reference
and technical significance within a system- or "theory3 '-language. We
summarily consider a few of the findings within these terms.

a. "Presystematic'5 analysis of experience. No one will dispute that
experiential analysis in this sense is a necessary condition to the psy-
chological enterprise. Nor has this point ever been disputed, even though
at times it has not been widely advertised. As against the temper of the
past few decades, however, we can say that there is evident in the thinking
of Hebb and Kelson (Vol. 1); Miller and Guthrie (Vol. 2); and
certainly in the "insightful anthropomorphism" of ethologists like Hinde
and comparative psychologists like Harlow (Vol. 2) a more far-ranging,
sensitive, and explicit dependence on presystematic phenomenology than
has been characteristic of earlier Age of Theory practice. Experience,
and in general the phenomena and involvements of human life have
been utilized as the matrix of problem and hypothesis formation, of initial
estimates for the plausibility of assumptions, etc., in a more direct and
less apologetic sense than has been usual.

b. Some transitional cases. Several authors in the study may be
characterized as holding an attitude towards "experiential analysis"
which falls somewhere between its presystematic and its systematic use.
The most instructive case is that of Tolman who, as we have already
seen in the discussion of intervening variables, now expresses the de-
pendence of his theorizing on his own phenomenology in a way which
makes the objectivist nuances of his theory language broadly meta-
phorical. The clear statements to this effect contained in the extended
quotations from Tolman towards the beginning of this epilogue (pp.
736-737) are echoed at many places in his essay. Where it is not within
the lines it is between them. In fact, after reading Tolman's presentation,
it is more difficult than ever to avoid the impression that much of the
power of his thinking derives precisely from his use of, or fidelity to,
"common sense" conceptual categories of a cognitive sort. One feels
more strongly than ever that whatever the inadequacy of the assumptions
made in such a vocabulary, they will probably be in some sense less
"vicious" than assumptions which, because of a principled commitment
to some simplistic vocabulary, are forced into abusing and distorting
ontology.

Also in this transitional group are certain of the students of the
person and the social setting whose tendency to hold on to the language
of "behavior" and associated imagery is rendered obsolete by the nature
of their problems, their predictive aims, the character of their variables,

768 SIGMUND KOCH

and even certain of their expressed metatheoretical attitudes. Time after
time among these systematists we encounter circumlocutions and euphe-
misms which are being made increasingly unnecessary by the implications
of their own work.

c. "Systematic phenomenology.53 Here we merely acknowledge that
men like Gibson (Vol. 1), Prentice (representing Kohler, VoL 1),
Cartwright (representing Lewin, Vol. 2), Asch, Murray, and Rogers
(Vol. 3), all regard variables having direct experiential reference as
legitimate elements of systematic analysis. And most systematists in
this group explicitly introduce such variables. Some among this number,
like Gibson, confine the role of experiential variables to the systematic-
dependent side of their relationships. Others, like Prentice in his analysis
of Kohler's theory, regard "phenomenal" variables as either dependent
or intervening but, at least by implication, proscribe them from the
systematic-independent category. Still others, like Murray, are willing
to contemplate the utilization of experiential variables at any position in
the systematic array.

In general, though, it can be said that whatever the attitude taken
towards experiential analysis, there has been no marked tendency among
authors in the present study (with the single exception of Henry
Murray) to join in any explicit way the many methodological and
empirical questions that might be asked concerning fruitful and rigorous
utilization of experiential data. Certainly the manifold current stresses
against behaviorist epistemology invite such questioning. Yet issues con-
cerning optimal techniques for experiential observation, the formulation
of adequate dependent variable categories, the integration of behavioral
and experiential data, the construction of theoretical concepts from ex-
periential data, etc., have been addressed by indirection, if at all.

Summary re Observation Base

In this section on the observation base and its relation to the end-
terms of systematic analysis, we have laid before the reader several trends
which seem to define a growing stress — both internal and extrinsic —
against behaviorist epistemology. Since, during the Age of Theory,
certain core assumptions of this epistemology have determined the reign-
ing conception of the observation base of psychological science, that con-
ception seems called into question. As we have already seen, this same
conception is sharply challenged by those findings of the study discussed
in connection with the "unambiguous linkage" criterion of the inter-
vening variable paradigm. The current analyses of S and R at the level
of molar behavior theory make it fairly safe to say that use of these end-
terms of systematic analysis has involved no guarantees of empirical
significance or meaningfulness. If stimuli and responses are acknowledged

Epilogue 769

to depend for their identification on the perceptual sensitivities of human
observers, then the demand for something tantamount to a language of
pointer readings, whether as simple energy-source or movement descrip-
tions, or as disjunctions of fixed stimulus "indicators" and response
"measures/3 must be given up. And if this demand be given up, then
much time-worn argumentation as to the intrinsic ambiguity of an ex-
periential language, or in fact any language the end-terms of systematic
analysis in which are not S and R, becomes idle and beside the point.
If, further, the requirement is asserted that S be specified in a way which
includes its inferred meaning for the organism, then any basis for a
difference in epistemological status between an S-R language and what
has been called "subjectivistic" language is eliminated. There may be
objectivistic paradigms for the "representation" of meanings in certain
simple cases; there is no behavioristic paradigm for their determination
in most of the human cases that count.

If we consider the trends connected with S and R and add to them
the other trends discussed in this section, we may conclude that no one
of the descriptive patois available to psychology at the present time has
a privileged status with regard to semantic significance — on any criterion
of meaning. From the point of view of semantic purity or innocence,
we can just as well talk a "crypto-objectivistic" cognitive language a la
Tolman, an experiential language a la Murray, or a hypothetical con-
struct language a la Hebb as we can a homogeneous language of S and
R. The trends in this section, along with those in earlier sections, suggest
there to be grave defects in the analyses of empirical significance that
have ruled psychology for the past several decades. A re-examination of
the theory of definition of a sort adjusted to the realities of practice and
the demonstrable business of psychology seems clearly indicated.

IV. MATHEMATIZATION OF SYSTEMATIC RELATIONSHIPS

Perhaps the most passionate Age of Theory demand has been that
for the mathematization of systematic relationships — preferably at levels
of quantitative specificity at least comparable to classical physics. Espe-
cially during the mid-phase of the Age of Theory (say 1935-1945), it
seemed to most a matter of course that the goal of science, and thus of
psychological science, was over its entire range the statement of mathe-
matical laws. The reconstruction of theoretical practice in physics which
governed Age of Theory ideology seemed to make attainment of this
goal a matter of destiny — in the expectation of many, rather short-range
destiny.

In striking contrast to this requirement, it is a fact that throughout
the early and classic Age of Theory, only a narrow range of mathe-

770 SIGMUND KOCH

matical strategies was tried out (at least at the level of theoretical rela-
tionships having reasonably general intent). For the most part these
were modeled quite literally on the use of mathematics in certain of the
simpler contexts of classical physics — usages ranging from simple algebra
and analytic geometry to application of differential equations. Lewin's
programmatic use of topology might be noted as a lonely exception.
The strategies ran a rather narrow gamut from the rational assumption
of basic laws of learning (e.g., such relatively early Age of Theory de-
velopments as the rational learning analyses of Thurstone, Gulliksen
and Wolfle, Woodrow, etc.), through attempts to build up descriptions
of functional relationships via empirical curve-fitting techniques (the
early Hull), to methods representing some combination of such "rational"
and "empirical" ingredients (e.g., the later Hull). In retrospect, it seems
also fair to say that the absolute volume of effort in these directions was
small.

Indeed, one can only conclude that the strength of the Age of Theory
autism for quantification led to some blurring as between aim and
achievement. A kind of pseudo-mathematical jargon became common to
all. This is indicated in the general fondness for the language of
"variables" and "functions," the incessant use of terms like "parameter"
and "parametric" in purely metaphorical contexts, etc. Moreover, the
intervening variable schema proved a ready milieu for facile talk in this
idiom of wish. The ideal case of the "explicit and determinate" con-
struct linkages called for by the intervening variable paradigm is, of course,
precise quantitative specification. When the assumptions of the sys-
tematist were not put forward in apparent quantitative form, there
appeared almost always an aside representing them as transitional, first-
approximation statements which would give way to precise mathe-
matical specification with the inexorable advance of the given theoretical
program.

The trends of the present study are in distinct contrast to the state
of affairs above described. Relative to the mid-Age of Theory, there is
a marked increase in realism concerning the prospects for strong degrees
of mathematization, especially in formulations having relatively general
systematic objectives. There is more modesty, contextualism, and grad-
ualism in representing accomplishments, acknowledging difficulties, and
estimating feasibilities. These attitudes gain all the more force in that
they are correlated with the (well known) marked increased in recent
years in the range of mathematical strategies tried, and in the volume
of mathernatico-theoretical effort.

We proceed to a brief sampling of the attitudes towards the status
and prospects of mathematization in psychology of two echelons of

Epilogue 771

systematists: (1) individuals who make no marked use of mathematical
procedures at systematic levels, and (2) individuals who do.

1. "Non-mathematical" Systematists

The attitude of the authors in this group is well represented by
Tolman's simple statement: "I am very relaxed about this" (Vol. 2,
p. 97). Moreover, there would be general agreement when he says:

Pyschology today seems to me to be carried away (because, perhaps, of
feelings of "insecurity") into a flight into too much statistics and too great
a mathematization. . . . But to me, the journals seem to be full of overso-
phisticated mathematical treatments of data which are in themselves of
little intrinsic interest and of silly little findings which, by a high-powered
statistics, can be proved to contradict the null hypothesis (p. 150) .

Indeed there would be wide sympathy with Tolman's further comment
when specifically addressing the issue of quantitative specification of
function forms:

As I have already indicated above, such attempted precision seems to me
for the most part premature. Wherever one can do it, the experimental
conditions are so overcontrolled, restricted, and specific that any valid
generalizations from such attempts seem to me impossible (p. 150).

Similar sentiments have been expressed in a variety of ways by Hebb
and Morgan (Vol. 1); Guthrie, Hinde, Miller, and Skinner (Vol. 2);
and Rapaport, Rogers, and Thelen (Vol. 3). And the same position is
clearly implicit in the presentations of such men as Asch, Katz and Stot-
land, Murray, and Newcomb (Vol. 3). These systernatists should cer-
tainly not be construed as anti-quantitative: they are imposing no re-
strictions on the future but would, with varying emphases, agree with
Hebb's statement that "Precise quantification with respect to theoretical
entities should be expected only in late stages of development of the
science" (p. 636).

Certain points recur with high frequency among members of this
group. Thus, the general caution, evident in the Tolman quotation
above, to the effect that there is often (and perhaps in principle) an
inverse relation between quantitative specificity and empirical generality
in psychological statements is often registered. There is also a tendency
to rediscover the existence of certain respectable but non-quantitative
sciences, or phases of given sciences, which had more or less dropped
from view during the classical Age of Theory. The classificatory and
descriptive branches of biology are often pointed to, as is evolution the-
ory in its pre-mathematical forms. Meteorology comes under notice as a
mixed case. A related point often made concerns the interpenetration of
qualitative and quantitative analysis, even in the most highly quantita-

772 SIGMUND KOCH

tive fields. People are beginning to remember with Hebb "how often it
was the development of a new [substantive] idea that made quantifica-
tion possible thereafter" (p. 636).

Such attitudes and judgments as have been sampled are not new.
Many of the men who hold them now held them ten or even twenty
years ago. What is new is the direct, non-apologetic, and sometimes even
truculent way in which they are expressed. It is also instructive to note
something which could have only become evident in a broadly deployed
study like Study I — the wide spread of the present agreements across
differences in conceptual predilection and problematic interest.

2. Systematists Working towards "Strong" Degrees of Mathe-
matization

The diversification of mathematical strategies and increased math-
ematical effort characteristic of recent years is well documented by the
present study. The contributions of Blank, Licklider, Graham, Pirenne
and Marriott, and Helson (Vol. 1); Estes, Ellson, Frick, and Logan
(Vol. 2); Cattell, and Lazarsfeld (Vol. 3), among others, give an
illuminating sampling of the range of mathematical imagination that
has been shown. These men show precisely that dedication to their
methods that one would expect from creative scientists. And certainly,
in varying measures, they are optimistic about the potentialities of
their methods. Yet their general estimates as to present achievements,
prospects, and limiting possibilities for mathematization in psychology
are not much less conservative than those of the systematists discussed
above. Relative to earlier Age of Theory doctrine, there is a remarkable
increase in the disposition to define boundaries and point up limits.
There is also a realistic concern with many knotty problems concerning
the preconditions to significant mathematization, problems many of
which were wishfully bypassed in earlier Age of Theory thinking.

A few examples of the current realism and contextualism among
mathematically oriented psychologists may be of interest.

Those individuals who represent the position that probably did
more in the Age of Theory to foster quantitative optimism than any
other now estimate achievement and prospect in subdued terms. We
have already seen that Miller sharply rejects "the misleading trappings
of pseudo-quantification.53 Though he is "painfully aware of the dis-
advantages of a qualitative theory limited to ordinal scales and pre-
dictions of 'greater than'," he nevertheless sees "some virtue in the
strategy of putting one's theoretical notions through qualitative tests
first" (Vol. 2, p. 281). Unlike many earlier workers in the Hullian
tradition. Miller is critically sensitized to the indeterminacy of units
in the ordinal scales used for most behavior measures, and thus by im-

Epilogue 773

plication the indeterminacy conferred upon function specifications at
mathematical levels which presuppose stronger orders of measurement
as a condition to significant empirical application. Though Logan, in
representing Spence's position, does not depart so far from tradition as
does Miller, the attitude towards quantification is certainly a gradual-
istic one. What in basic conception is still a Hull-type quantification
program is seen as central to Spence's effort, but it is made clear that
quantitative analysis is rather conservatively phased in with the de-
velopment of relevant data, and that the resulting quantitative func-
tions are so formulated (or conceived) as not to overlap their defining
base by untoward amounts.

Turning to another quantitative program laid down during the
Age of Theory, classic phase, there is certainly no tendency in Cart-
wright to overstate the significance of Lewin's use of topology (along
with such modifications as are represented in his "hodology") — neither
in a technical mathematical sense nor a psychological one. This would
be clear if only from the candor with which Cartwright exposes the
difficulties in conceptualizing Lewin's basic notion of the "life space"
(Vol. 2, pp. 65-72). Moreover, in reporting the recent work of the
Michigan group designed to represent relations formerly described by
Lewinian topology in terms of linear graphs, a measured tone is main-
tained wholly uncharacteristic of the earlier Age of Theory. While on
this topic it is well to note that whatever the ultimate fate of Lewin's
mathematical notions, he had an early priority among general psycho-
logical theorists in choosing areas of mathematics for psychological ex-
ploitation more on contextual grounds relative to apparent problematic
requirements than on grounds of emulating practice in physical science.

Information and (to a lesser extent) servo-mechanism theory have
sometimes in recent years been associated with a type of free-roving
optimism not dissimilar to that of certain earlier Age of Theory pro-
grams. And indeed, each of these formulations has developed its own
flexible analogical patois. Yet, readers of Frick on information theory
and of Ellson on linear frequency theory (Vol. 2) will come away with
no impression that these formulations exhaust the universe of mathe-
matical analysis relative to psychological phenomena.

They will find Frick emphasizing that substantive exploitations of
information theory have to date been thinly scattered and must by the
mathematical restrictions of the formulation be limited to rather nar-
row modes and levels of analysis. They will discover that many who
have "applied" information theory have in the first instance misin-
terpreted what is essentially a formal mathematical system for a "sub-
stantive model" and have thus, "for instance, been able to confound
thermodynamics and the statistical structure of language" (Vol. 2, p.

774 SIGMUND KOCH

612). They will discover, further, that "the formalism of the theory is
directed at a determination of the efficiency of communication, and the
application of information theory to psychological data implies an in-
terest in the efficiency, rather than the structure, of the process under
study" (p. 613). And they will hear it said more forcefully than is
usual that information theory, like all probability models, calls for
strong simplifying assumptions with respect to the data to which its
methods of analysis are applied — assumptions rarely approximated in
fact. Within such limits, Frick develops cogent, if measured, grounds for
the fruitfulness of information analysis as applied, e.g., to certain as-
pects of sequential behavior and behavior patterning, while Licklider
(Vol. 1) gives admirable documentation of the fruitfulness of informa-
tion analysis in certain aspects of auditory theory.

The case for linear frequency theory is presented with comparable
sobriety by Ellson (Vol. 2). Though hopeful for the ultimate prospects,
he makes it entirely clear that not in a single study of human behavior
has the central condition for the applicability of linear frequency anal-
ysis— the criterion of linearity — been met.

The expanding class of stochastic models for the systematization of
learning is well represented by Estes (Vol. 2) in the presentation of his
theory. Though Estes has a careful program for the extension of his
formulation from its base in the description of simple acquisition func-
tions for lever pressing, he is equally careful to circumscribe the senses
in which he seeks generality and the ranges within which it has been
achieved. He is also very explicit in defining the simplifying assump-
tions of his model and the tight restrictions in problem formulation,
independent and dependent variable characteristics, experimental de-
sign, etc., necessary even for their approximate satisfaction.

A most interesting departure from earlier Age of Theory mathe-
matical strategy may be found in the work of Lazarsfeld (Vol. 3).
Earlier Age of Theory thinking had seen the problems of scaling and
(quantitative) theoretical construct formation as relatively independent
problems. Scaling methods developed in such contexts as psychophysics,
test theory, and attitude measurement, while efforts towards quantita-
tive construct inference proceeded in the hands of learning and "be-
havior" theorists. Against this background, it is of interest to see the
scaling theorist, Lazarsfeld, making ingenious efforts towards the ex-
tension of a mathematically sophisticated scaling method to problems
of theoretical concept formation. The contrast with learning theory is
instructive in that Lazarsfeld joins the problem of social science theory
at a far more "primitive" level — that of initial concept formation. Un-
like typical Age of Theory practice, the question is not one of establish-

Epilogue 775

ing mathematical construct linkages, but merely one of establishing con-
structs which could be linked (or prove "linkable'3) in future theoretical
analysis. Though there is a certain sweep in Lazarsfeld's programmatic
extension of latent structure analysis to the general problems of concept
formation in all social science, there is also, as everywhere among the
present group of authors, no tendency to suppress difficulties or overstate
achievements. Basically, his concern is to open up a line of methodo-
logical speculation which has been bypassed in the rush towards "high
order33 theory. The procedures are meant to "clarify how we create
'underlying3 concepts like traits, attitudes, group characteristics, etc.33
whose "role is to summarize a variety of empirical observations and to
store them, one might say, for systematic use in a 'theory3 which we hope
will one day develop33 (Vol. 3, p. 485) ; moreover, it is "difficult to pre-
dict whether LSA will be useful if applied to conceptually more complex
intervening variables as they appear, e.g., in learning theory33 (p. 537) .

Not inconsonant with the gradualistic overtone of Lazarsfeld3s anal-
ysis are certain points in CattelFs discussion of factor analysis (Vol. 3).
That Cattell makes strong claims for the utility of factor analytic meth-
ods is more than slightly evident, but there is certainly no suggestion that
these methods preempt other mathematical approaches. The range of
application is clearly restricted to the initial isolation and identification of
variables and there are definite implications to the effect that factorial
methods are not relevant to theoretical analysis in any ultimate sense.

Finally, as Conrad Mueller notes in his supplement on sensory theory,
the sensory psychologists in this study, while reviewing their own work
and that of others, illustrate the use of a very wide range of mathemat-
ical methods in specific theoretical contexts. Indeed, the diversity and
richness of mathematico-theoretical strategies in this area are greater than
in any field of psychology. Whether such methods are extensible to areas
in which independent and dependent variables are epistemically more
complex is, of course, uncertain. And whether comparable success can
be expected from other mathematical methods in such areas is also un-
certain. Here as before, the most useful export from the sensory area
could be a rather general lesson — that of contextualism. It is interesting
to contemplate that during the classic Age of Theory when hopes were
large for the discovery of some canonical quantitative method adequate
to all problems of psychology, the repository of mathematical experience
accumulated by sensory psychology was rarely consulted. Instead, the
methodology of the day consulted the history of other sciences rather
than its own. If what we have been calling "contextualism33 was then
too painful a lesson to bear, there are many indications that this is so no
longer.

776 SIGMUND KOCH

V. FORMALIZATION AND PSYCHOLOGY

Presupposed by and regulating all elements of Age of Theory ideol-
ogy is, of course, the hypothetico-deductive model of scientific theory.
The relevant issues are so widely ramified that in a sense we have been
discussing them all along. Yet precisely because of this dense texture of
ramifications, the understanding of recent psychological history requires
careful direct consideration of the role of the hypothetico-deductive
model. Here again we must relinquish adequate consideration for scat-
tered generalizations.

It is well to start with a distinction — that between the "hypothetico-
deductive model" and the c 'hypothetico-deductive prescription" The
hypothetico-deductive model is an epistemological reconstruction of the-
oretical method in science. It is a technical reconstruction based upon a
long tradition of work in logic, the philosophy of science, and general
epistemology. Its codification in the logical positivism of the late 'twenties
and early 'thirties was perhaps the clearest response to certain of the
relevant questions until that time. Though based only on selected for-
mulations of a relatively advanced character in classical and modern
physics, this codification had great technical importance, in that it
seemed to mark the beginning of an adequate answer to a central prob-
lem plaguing the history of scientific philosophy — that of how to state the
relationship between the rational and empirical components of science.
On the other hand, the hypothetico-deductive prescription would be the
recommendation, in any given scientific context, that inquiry be regu-
lated by the immediate aim of hypothetico-deductive systematization and
that the results of inquiry be set in correspondence with the explicit re-
quirements of the hypothetico-deductive (i.e., axiomatic or postula-
tional) model. We should notice at once that the plausibility of the
hypothetico-deductive model as a reconstruction of practice in given
areas of science, or even its desirability as an ultimate goal in various
sectors of the scientific enterprise, says nothing about the general feasi-
bility, or even fruitfulness, of the hypothetico-deductive prescription at
all phases of a given science, in all areas of that science, or, indeed, in
all sciences.

It is understandable that the molders of the Age of Theory — so
ardently in search of the "rule" of theory, of a decision procedure which
might contain assurance of forward movement — interpreted the hypo-
thetico-deductive model as the prescription that explicit axiomatic meth-
ods be applied "here and now" in psychological and social science. In-
deed, this interpretation, in its more austere versions, held failure to con-
form to some explicit pattern of formalization as a mark of obscurantism
and a confession of conceptual bankruptcy.

Epilogue 777

As in the (closely related) case of the mathematical ideology of the
Age of Theory, it is here necessary to note a similar discrepancy between
prescription and practice. To be sure, through the mid-thirties to the
mid-forties there were attempts, sometimes laborious ones, within in-
fluential theories to approximate the forms of hypothetico-deductive pro-
cedure. These ranged from the relatively informal "derivation" in nat-
ural language of "theorem" sequences from qualitatively stated postu-
lates to use of the combined resources of symbolic logic and mathematical
notation in the axiomatic treatment of limited ranges of data. But again
the absolute volume of such effort was small. Instead, the atmosphere
became permeated with an "imagery" of hypothetico-deduction — the
use or presence of which often seemed interpreted as equivalent to hypo-
thetico-deductive practice. A language of "postulates," "derivations,"
"primitive terms," "defined terms" (and in more ratified cases, logical
variables, constants, arguments, predicates, operators, functors and con-
nectives) became the tongue of psychological commerce. Since the more
powerful forms of hypothetico-deductive systematization involve quanti-
tative postulates which may then be manipulated by appropriate math-
ematical rules of inference, this language was, of course, one with the
mathematical language of "functions," "equations," "variables," "con-
stants," "parameters," etc., that we previously sampled. And of course
the language of operational definition and that of the intervening var-
iable schema as it developed in the rnid-thirties and later also finds its
place in the imagery system of hypothetico-deductive method.

The use of this imagery was not merely decorative and idle. The
acceptance of the hypothetico-deductive prescription had important con-
sequences for the prevailing conception of the aims of psychology, the
conception of where psychology stood in relation to its aims, and thus
the indicated route for further progress. It was, for instance, assumed by
many that a backlog of significant empirical knowledge existed adequate
to the "construction" of broad-scope, if not comprehensive, theories con-
forming to the requirements of the hypothetico-deductive model. It was
believed that psychology was at a stage such that theoretical differences
would inevitably and almost automatically be resolved by the "differen-
tial test" of "derivations" from rival "postulate sets." Perhaps of most
serious import for the character of actual practice was a cluster of be-
liefs to the effect that adoption of the forms of the hypothetico-deductive
model (or the imagery of its forms) guaranteed that the scientific enter-
prise would be "self-corrective." Such beliefs led, for instance, to the
strange expectation that the initial plausibility of a "postulate" is of
little moment in that proper adherence to the forms of hypothetico-
deductive method would almost certainly refine its adequacy or lead to
its early demise.

778

SIGMUND KOCH

Once more we may report that the trends of the study are in definite
contrast to the earlier Age of Theory position. Here, too, a more gradual-
istic and contextual attitude is shown in delineating achievement and
prospect. Few authors in this study would "scrap" the hypothetico-
deductive model as the stipulation of a methodological ideal, ultimate ap-
proximation of which would be highly attractive. Most, however, would
challenge the feasibility of the hypothetico-deductive prescription (in the
sense of any highly explicit or "strong" axiomatization) as an immediate
program for all domains of systematic effort, or indeed for any systematic
enterprise contemplating reasonably broad empirical reference. This
challenge is, of course, delivered with different force and on different
grounds by different men. It certainly cannot be said that the imagery
of the hypothetico-deductive prescription no longer has effects — espe-
cially those more indirect ones which determine problem selection and
modes of problem formulation, and color the content and statement of
programs for inquiry. Nor can it be said that there is not frequent dis-
comfort, sometimes guilt, over the inability to claim a more advanced
status against a hypothetico-deductive "measure." But the hypothetico-
deductive prescription has lost much of its force.

In sampling attitudes, we roughly distinguish three positions. If the
reader finds that each step on our "scale" is characterized in some com-
plexity, he may be assured that this is as nothing to the quiddities of
individual positions which, for obvious historical reasons, are intricately
stratified in this area.

1. Belief That the Hypothetico-Deductive Model Represents Scien-
tific Practice in an Incomplete and Possibly Misleading Way; Convic-
tion That the Hypothetico-Deductive Prescription Is Infeasible

This position is perhaps most fully documented by Skinner, whose
entire essay may be interpreted as a reaction to hypothetico-deductive
prescriptionism. The incidence of his critique is well conveyed by the
following generalizations about his own scientific behavior:

The notes, data, and publications which I have examined do not show
that I ever behaved in the manner of Man Thinking as described by John
Stuart Mill or John Dewey or as in reconstructions of scientific behavior
by other philosophers of science. I never faced a Problem which was more
than the eternal problem of finding order. I never attacked a problem by
constructing a Hypothesis. I never deduced Theorems or submitted them to
Experimental Check. So far as I can see, I had no preconceived Model of
behavior — certainly not a physiological or mentalistic one, and I believe,
not a conceptual one. The "reflex reserve" was an abortive, though opera-
tional, concept which was retracted a year or so after publication ... It
lived up to my opinion of theories in general by proving utterly worthless

Epilogue 779

in suggesting further experiments. Of course, I was working on a basic
Assumption — that there was order in behavior if I could only discover it —
but such an assumption is not to be confused with the hypotheses of deduc-
tive theory (Vol. 2, p. 369) .

From a quite different incidence, Guthrie challenges both the feasi-
bility and the sense of the hypotheticodeductive prescription :

The fact that it had taken Russell and Whitehead some 400 pages to
establish the conclusion that one plus one equals two, and that every in-
tervening step could be challenged and would require more proof, and that
the steps of these added proofs would require still more, has made me im-
patient with the notion that there can be any completely rigorous deduc-
tion, or ultimate validity in an argument This scepticism colors my notions
of the nature of scientific facts and scientific theory (Vol. 2, p. 161).

Other important views of Guthrie concerning this issue are scattered
through his essay. Here we might add another of his points — one which
cautions against premature formalization in a way echoed by many other
contributors:

It will be a very long time before we are prepared to formalize our ac-
count. The problem of reinforcement vs. contiguity should be settled before
embarking on a system. When the choice is made too early, and the funda-
mental definitions and categories become official and items that all gradu-
ate students must master for the purpose of nationwide examinations, we
may find ourselves committed to unproductive efforts (p. 193) .

Tolman also declares himself a member of this group in these quite
definite words:

All I can say here is that my system is based on hunches and on com-
mon-sense knowledge. It is certainly not "hypothetico-deductive." I have
not the type of mind that can remember which were my axioms and which
were my deductions. In any event, if a system were a SYSTEM, which I do
not believe psychology to be, it would be largely arbitrary which one took
as axioms and which one took as derivations. To attempt to build psychology
on the analogy of a closed mathematical or logical system seems to me a
"bad error" (Vol. 2, p. 150).

Though it is rare that other authors in this group express themselves
with the same lack of ambivalence as do those just cited, many make
points of the same order. Thus, for instance, Rapaport, who generally
stresses the need for the systematic tightening of psychoanalysis, in an
excellent brief discussion of "The desirable level of formalization" says:

Actually, axiomatization has always been a late product in every sci-
ence. Centuries of Egyptian geometry preceded Euclid. Newton had not
only Galileo and Kepler, but thousands of years of physics behind him.

780 SIGMUND KOCH

Sciences do not arise from, but culminate in, axiomatics. Axiomatic systems
do not reveal the tracks of a science's development; they conceal them
(Vol. 3, p. 135).

Several men in the study who do not directly address the isssue of
"formal organization" show by the entire trend of their essays and by
positions expressed on other matters that they would fall into the present
group. Murray, Asch, and Thelen (Vol. 3) clearly would, as would
probably Gibson (Vol. 1).

2. Belief That Formalization Is Desirable in Short-range Future,
but Strong Awareness of "Dangers" and Difficulties

Cart wright's observations are of particular interest in that the mid-
Age of Theory Lewinian tendency was to warn against the "freezing
effect" of "premature formalization," but nevertheless to flirt with the
possibilities of rendering components of the theory in qualitative-verbal
"axiomatic" form. The "warning" now comes through in more resolute
terms than formerly:

Contemporary psychological theory is in danger of losing touch with
empirical reality. The placing of too great a value upon formal elegance in
the construction of theory may well create an insurmountable chasm be-
tween the theorist and the psychologist who is interested in the naturally
occurring behavior of people. Unless future interest in "model building"
is closely guided by an unbiased reference to empirical facts, formal ele-
gance will be purchased at the cost of empirical applicability. . . . While
the traditional emphasis of Lewinian psychologists upon the hypothetico-
deductive method is still appropriate, one should not forget that formal
theory is useful in an empirical science only in so far as it serves as an aid
to description.

. . . Too much of current psychological research, I fear, is designed
not- so much to discover new facts as to confirm some derivation from a
limited formal theory. It is in the formulation of research problems that
the major advances of psychology take place. Careful observation, record-
ing, and measurement of naturally occurring events and of "experiments of
nature" will for a long time to come be the most important source of the
significant problems of psychology. For this reason, it would seem wise for
psychologists to avoid any premature judgment that painstaking observation
is inferior or antithetical to rigorous theory (Vol. 2, pp. 80-81) .

3. Demonstration of Some Degree of Achieved Axiomatic Explicit-
ness in a Limited Area, plus Measured Optimism over the Prospects
for Extension at Comparable Levels of Axiomatization

A highly mixed group of individuals could be said to take this posi-
tion in one way or another. These would include the sensory psychologists,

Epilogue 781

those concerned with limited-scope mathematical models (whether pri-
marily substantive or "methodic35 ), and people like Miller and Logan
who, in subdued form, present a picture of theoretical method rather
closer to dominant mid- Age of Theory conceptions than most other au-
thors in the study. Heterogeneous as this group is, certain important
generalizations can be made. Thus, for instance, none of these men give
evidence of an overweening commitment to axiomatic method as an end
in itself, nor do most of them go out of their way to generate optimism
about the general feasibility, or even fruitfulness, of formalization in areas
outside the limited context in which they work.

Licldider well illustrates the characteristic problem-centeredness of
the sensory worker (with respect to this issue as elsewhere) in indicating
the basis of his choice of an "analogue level55 rather than an axiomatic
mode of formulation. Thus he points out:

My own experience in thinking about auditory problems leads me to
doubt that a highly formal axiomatic approach would be very helpful to
me at the present stage. On the other hand, informal exposition (of the
present kind) fills up a great amount of space if the problem is com-
plex ... As a compromise, it may be convenient to think of the auditory
process as a system of operations upon variables in approximately the way
that analogue computer experts visualize their computational problems. . . .

Between the axiomatic and the analogue levels, there is, I believe, com-
plete translatability. The axiomatic level is better for examining theories as
theories. . . , The analogue level is better matched to most people's ordi-
nary modes of thought and is, therefore, likely to facilitate interactions be-
tween theory and experiment. . . .

In expressing a mild preference for the intermediate or analogue level
of formulation over the axiomatic, I am suggesting only that auditory
theory is in a formative stage and will probably not soon mature. There is
more need to line the theories up with the facts than there is to state them
in esthetically pleasing form. Even in mathematical logic, it appears, the
road to understanding involves processes of thought quite different from
those that are reflected in the final efficient, consistent, step-by-step deduc-
tion from postulates (Vol. 1, pp. 50-51) .

It will, of course, be evident from the presentations of Frick and
Ellson (Vol. 2) that as formal mathematical systems information theory
and linear frequency theory can be axiomatized with considerable rigor.
Frick, however, takes the position that at substantive levels, information
theory primarily (a) specifies a technique for data analysis, and (b)
"narrows33 the universe within which the systematist may construct em-
pirical models in which information analysis may be used. Thus, there is
no necessary carry-over from the axiomatic status of the formal theory
to that of any substantive model constructed with its "aid." Ellson, on

782

SIGMUND KOCH

the other hand, represents linear frequency theory as a model which
may in principle be given an empirical "interpretation" by "the addition
of a few empirical definitions which assert equivalence between (a) em-
pirical operations and (b) terms and logical operations in the model or
in statements derived from it" (p. 657). Nevertheless, as Ellson makes
abundantly clear, no such interpretation for any sizeable range of be-
havioral data can be valid at the present time,

Estes takes a position having much of the ring of classic Age of
Theory ideology re f ormalization :

One frequently hears the argument that so long as an empirical science
is in a primarily exploratory stage, theories must be informal and qualita-
tive. I do not question that informal and qualitative theorizing is sometimes
necessary and even rewarding, but I do have doubts as to both the necessity
and the wisdom of being long satisfied with it. The disadvantage of permit-
ting a haze of ambiguity to cover an entire theory is that the theorist, like
anyone trying to navigate in a fog, can never really tell how far he has
come or whither he is heading. Although we cannot get rii of ambiguity
entirely, we can localize it by making our theoretical concepts and assump-
tions precise and permitting indeterminacy only in the correspondences be-
tween theoretical and empirical variables. . . .

It will be clear from our analysis of the present theory that all linkages
among constructs are explicit and determinate and all derivations of
theorems are accomplished by exact mathematical reasoning. Interpretive
rules, on the other hand, are somewhat open ended. . . .

This ring, however, itself somewhat muffled, is further subdued in the
following paragraph:

I would like to emphasize that my brief for rigor in theorizing does
not imply any great love of formality for its own sake. In the developmental
stages of a science it is not healthy for theories to stand still long enough
for exhaustive logical analysis. The kind of formalization I consider neces-
sary to sound theory construction consists in progressively sharpening the
definitions of concepts and exposing concealed assumptions at the same time
that the theory continues to undergo correction and refinement in the light
of experimental applications (Vol. 2, pp. 472-473).

A still more nearly classic Age of Theory emphasis is rather generally
evident in the essays of Miller and Logan (Vol. 2). Yet Miller, in de-
veloping his qualitative systematization of "conflict," is concerned mainly
with using this as a constructive device for isolating certain of the oft-
bypassed problems of systematic work. And Logan, in representing the
Spence position, certainly claims no high degree of axiomatic specificity,
nor does he regard it as "practicable at the present time to write a fully
formalized behavior theory of any general significance" (p. 329 ) .

Epilogue 783

A CONCLUDING PERSPECTIVE

It can in summary be said that the results of Study I set up a vast
attrition against virtually all elements of the Age of Theory code. If all
contributors are not eager to express their intransigence in neon script,
neither do they conceal their doubts and questionings. No one is pre-
pared to retreat one jot from the objectives and disciplines of scientific
inquiry, but most are inclined to re-examine reigning stereotypes about
the character of such objectives and disciplines. There is a longing, bred
on perception of the limits of recent history and nourished by boredom,
for psychology to embrace — by whatever means may prove feasible —
problems over which it is possible to feel intellectual passion. The more
adventurous ranges of our illimitable subject matter, so effectively re-
pressed or bypassed during recent decades, are no longer proscribed.

For the first time in its history, psychology seems ready — or almost
ready — to assess its goals and instrumentalities with primary reference to
its own indigenous problems. It seems ready to think contextually, freely,
and creatively about its own refractory subject matter, and to work its
way free from a dependence on simplistic theories of correct scientific
conduct. The day of role playing as a route to reassurance may be draw-
ing to a close. If our science cannot, in terms of attainment, feel secure,
it is at least the case that the dance of respectability, as called from the
wings by some fashionable theory of proper science, is no longer a de-
pendable source of security.

This preparedness to face the indigenous must be seen as no trivial
deflection in the line of history. It is well to bear in mind that the de-
pendence of the Age of Theory on prescription from extrinsic sources is
but the most recent chapter in a consistent story of such extrinsic deter-
mination of ends and means :

The institutionalization of each new field of science in the early mod-
ern period was a fait accompli of an emerging substructure in the tissue
of scientific knowledge. Sciences won their way to independence, and
ultimately institutional status, by achieving enough knowledge to become
sciences. But, at the time of its inception, psychology was unique in the
extent to which its institutionalization preceded its content and its meth-
ods preceded its problems. If there are keys to history, this statement is
surely a key to the brief history of our science. Never had a group of
thinkers been given so sharply specified an invitation to create. Never
had inquiring men been so harried by social need, cultural optimism,
extrinsic prescription, the advance scheduling of ways and means, the
shining success story of the older sciences.

The "scientism" that many see and some decry in recent psychology
was thus with it from the start. It was conferred by the timing of its

784 SIGMUND KOCH

institutionalization. If psychology had been born a century, three cen-
turies earlier, it would have been less "scientistic." There would have
been that much less science, and science-of-science, to emulate. Those
who use the term "scientism" dismissively are sensing a problem but
decrying the inevitable. Yet, few who fairly look at the brief history of
our science could agree that the balance between extrinsically defined
tradition and creative innovation — prescription and production — has for
any sizeable interval been optimal. From the earliest days of the experi-
mental pioneers, man's stipulation that psychology be adequate to sci-
ence outweighed his commitment that it be adequate to man. From the
beginning, some pooled image of the form of science was dominant:
respectability held more glamour than insight, caution than curiosity,
feasibility than fidelity or fruitfulness. A curious consequence — even in
the early days when such trends were qualified by youth — was the ever-
widening estrangement between the scientific makers of human science
and the humanistic explorers of the content of man. It is, for instance,
significant that a Freud, when he arrived, did not emerge from the lab-
oratories of 19th century experimental psychology; nor was the ensuing
tradition of work particularly hospitable to his ideas until rendered des-
perate by the human vacuum in its own content.

The history of psychology, then, is very much a history of changing
views, doctrines, images about w hat to emulate in the natural sciences — •
especially physics. In the 19th century, this meant the extension of ex-
perimental method to subjective phenomena; for early behaviorism, it
meant the use of experimental method exactly as in physics (objectively) .
By the late 'twenties, there was much objective experimentation but few
bodies of clearly stated predictive principles comparable to the crowning
achievements of physics: its theories (e.g., Newtonian mechanics, rela-
tivity theory). Instead, experimentation sometimes seemed aimless, "the-
oretical" hypotheses but loosely related to data, and debate idle. We thus
get, beginning around 1930, the emulation of natural science theoretical
method. If the resulting Age of Theory soon tended to subordinate pur-
suit of the indigenous to the easier consummations of dependency on
extrinsic models, this was no new compromise.

It is anyone's guess as to whether we are still within the Age of The-
ory. This epilogue has barely suggested the scope of the attrition devel-
oped by the Study I analyses against the "reigning" image of systematic
practice; yet there is a tendency still to funnel activity through its con-
tours. Much of the attrition, though real, is still implicit in practice.
There is a new contextualism abroad, a new readiness to consider prob-
lem-centered curiosity a sufficient justification of inquiry, but much ef-
fort is still invested in apologetically reconciling such impulses with Age

Epilogue 785

of Theory code. Schedules have been re-defined; systematic claims local-
ized or, if general, made more modest; pre-theoretical knowledge has
found a higher priority in the economy; a wider range of subject matters
has begun to assert the right to autonomous systematic development; and
a wider variety of formulations has been granted "theoretical" citizen-
ship. But the images which govern positive systematic action are still, in
the main, Age of Theory images. Often when they do not govern action,
they serve as its rationalization. Despite the fact that action can only be
fully free when at peace with its presuppositions, there has been very
little direct effort towards the creative emendation of Age of Theory
doctrine.

Yet, the stresses against that doctrine not only are severe and far-
ranging, they are sufficiently clear and specific to show where creative
thought is needed, and even in some cases to point directions. Indeed,
one of the ways in which the substance of this epilogue can be read is
as a preliminary isolation of those contexts in which our contributors
seem most persuasively to call for a rectification of Age of Theory doc-
trine. Certainly no two students will agree in the diagnosis of such focal
contexts of questioning in all particulars. There are in fact a sufficient
number of convergences in this study to offer comparable weights of evi-
dence for many diagnoses. But it is difficult to believe that there will not
be fundamental overlaps.

One area, for instance, in which such an overlap would seem in-
evitable concerns the need for a theory of definition adequate to the de-
mands of psychology. Each of the five principal trend-areas reviewed in
this epilogue, and especially the two reviewed most extensively (i.e.,
"The intervening variable paradigm35 and "The observation base"),
converges on this issue. Indeed, it is probable that however one sliced
the trends of Study I, they would still converge on this issue. Any re-
centering of Age of Theory ideology which would truly liberate psy-
chology for confrontation of the indigenous must give high priority to
the many trying and subtle problems essential to a just understanding
of empirical definition, and its place in the systematic enterprise. To
develop such a suggestion further, however, would take us close to a type
of concern which, by our original definition, is not within the province
of this epilogue. Here it is meet merely to point up the need for sustained
and continuing effort in this direction, and to register the writer's inten-
tion to return to this theme in the concluding volume of the series.

Some who may be still inhabited by the autisrns of the Age of
Theory, if even residually, will no doubt tend to experience the tenor of
the findings here reported as depressing. To such persons it should be
urged that it is important to distinguish between the actual situation in

786 SIGMUND KOCH

psychology and certain of the traditional criteria against which its prog-
ress has been measured. The fact that the status may seem dim when
measured against such criteria as are contained in the intervening
variable paradigm., the demands for operational definition, for strong
orders of quantification and axiomatization, or indeed most other re-
quirements of the Age of Theory disenfranchises no positive accom-
plishment within psychology. Though it is possible to say that the direc-
tives implicit in Age of Theory doctrine have resulted in a constriction
of the range of interest, and perhaps some impoverishment within that
range, the important fact is that, as always in science and in problematic
action generally, there has been a definite gap between activity and its
rationale. That the contributors to this study have so sweepingly chal-
lenged Age of Theory values in terms of the actual tendency of their
own creative work is itself an indication of the size of that gap.

The fact that the maturity of our science has often been over-repre-
sented (relative to some extrinsic standard of scientific maturity) says
nothing with regard to the sound advances that have been made. If this
progress still mainly involves the search for fruitful variables, rather
than the finishing touches to elaborate general theories, it is at least the
case that certain of the initial skirmishes have been joined. What
emerges from the critique of Age of Theory ideology made by our
authors is a far more open and liberated conception of the task of psy-
chology, the role of its investigators and systernatists, than we have en-
joyed in recent history. There is refreshing recognition of the role of
creativity in all aspects of the scientific enterprise and a willingness to
confront the fact that creativity cannot be reduced to rule or scheduled.

In general, we are given reason to expect a widespread and profound
readjustment of rationale and action in our science. If the limits of
methodological, strategic, or programmatic thinking relative to their
constructive role in the development of science are now seen as tighter
ones, such thought is not therefore to be regarded as less important or
valuable. On the contrary, this circumstance makes "methodological"
thinking, and more generally the second-order analysis of inquiry, all
the more important. But if such thinking is to have fruitful con-
sequences, it must be directed at the realities of science. It must not in-
flate these realities into a myth of perfection, or indeed, impose some
facile myth of perfectibility upon science. Many concrete and extraor-
dinarily pressing questions are opened up by the specific gaps between
conventional methodological rationale and the realities of the systematic
endeavors represented in this study. It is in the creative confrontation of
these gaps, and in the readjustment of methodological depiction and
imaging to action that many important tasks for the immediate future

Epilogue 787

lie. Such readjustment cannot fail to have salutary consequences for
further action. Only from attempts to achieve this more sensitive ac-
commodation of rationale to action can there emerge a type of "metho-
dology"— a type of application of man's critical agency — which could be
of use to the practicing scientist.

Such has been the prestige in recent decades of the sources from
which psychology has derived its conception of the scientific process
that many individuals will perhaps be shocked at the strong implications
generated by our study concerning the inadequacies of that conception.
In fact, it is more or less inevitable that their sense of shock will be in
direct proportion to the clarity and force of those implications. It is no
secret, of course, that the primary source from which the Age of Theory
borrowed its notion of the scientific enterprise, especially at theoretical
levels, was the philosophy of science and, most directly, the logical posi-
tivism of the late 'twenties and early 'thirties. In this epilogue a special
attempt has been made to stay within the data of Study I and thus to
determine what might be learned about systematic practice in psy-
chology by studying systematic practice in psychology. Pursuit of the
indigenous is best advocated by precisely that pursuit. But at this point
it is well to stress that if there be any who are troubled by the discrep-
ancy between the results of this study and philosophical precept, their
worries are outdated. The trend of philosophical analysis for more than
twenty years, and conspicuously for the past ten, has been towards a
liberalization of precisely that view of theory which conditioned the
psychological Age of Theory — a liberalization which by now must be
adjudged immense. Contributing to this liberalization has been not only
logical positivism but such cognate movements as neo-pragmatism and
English analytic philosophy. It is of high significance to note that each
one of the trends of the present study which may seem so radical when
viewed against an Age of Theory measure is entirely consonant with the
newer philosophical views. Psychology, then, still bases its understanding
of vital questions of method on an extrinsic philosophy of science which
(in some areas) is twenty years or more out of date.

Consonance does not mean derivability; still less does it mean
identity. What psychology needs to know about its goals and stratagems
is far too subtly embedded in the tortuous quiddities of inquiring action
for the philosopher to be of appreciable help. It would, of course, be as
callow to maintain that philosophical analysis is of no relevance at all
as it would be to entrust philosophy with total responsibility for the map-
ping of our future. But the need for testing, culling, transforming, sup-
plementing, adapting philosophical insights within a context utterly con-
trolled by responsiveness to the indigenous is absolute. And, indeed, it is
probable that in the long run psychology will have more to contribute

788 SIGMUND KOCH

to central problems of epistemology and others traditionally associated
with philosophy than vice versa.11

This is only one of many epilogues that could have been written.
The five major topics selected for discussion seemed to offer the most
direct access to the significance of the study for history. Regrettably,
there has been little opportunity to discuss the fundamental substantive
contributions that have been made in virtually every essay — the changes
and refinements in established positions and the fertile new ideas that
have been generated. We have had no opportunity to discuss important
contentual convergences: the massive evidence of a tendency for what
were formerly discrete and rather insulated viewpoints to come closer, or
even merge in significant respects; the tendencies towards joint ac-
knowledgment of problems discriminated only by local groupings of in-
quirers in the past. All these matters, however, and many others are
open to inspection and collation by readers, each of whom will compose
the only kind of epilogue that can be truly meaningful.

Perhaps most discomforting of all is the fact that our thin and selec-
tive mobilization of results loses the true quality — the stimulation and
often excitement — of the individual essays. The ultimate import of this
study is to be found in no "trend," but in the fact of men speaking in
languages sufficiently robust to defy fusion.

"Volume 7 of this series — Psychology and the Human Agent — develops con-
crete grounds for this assertion. Also in that volume certain of the newer develop-
ments in the philosophy of science are considered, and an attempt is made to
specify the senses in which they are coherent with (yet far from equivalent with)
the trends of the present study. Many readers, of course, will already be apprized
of the directions (and they are not few in number) taken by the newer philosophy
of science in its liberalized reconstructions of the nature of theory, empirical and
Bother modes of definition, etc. Volumes I and II of the Minnesota Studies in the
Philosophy of Science (H. Feigl and M. Scriven (eds.), University of Minnesota
Press, Minneapolis, 1956 and 1958) register these developments admirably and
within a context addressed to questions of psychological methodology. Perhaps the
most dramatic index to the extent of the philosophical liberalization is Carnap's
article, "The Methodological Character of Theoretical Concepts," in Volume I,
which repudiates almost totally his earlier (1936-1937) analysis of empirical, and
thus operational, definition in terms of the "reduction sentence" — an analysis which
has dominated psychology ever since importation in the early 'forties.

SIGNIFICANCE OF
SENSORY PSYCHOLOGY

FOR CERTAIN

METHODOLOGICAL

PROBLEMS CONRAD G. MUELLER

Columbia University

Introduction 789

Some difficulties in the statement of methodological problems . . . 789

Some contributions of sensory psychology to methodological problems . 790

Some Characteristics of Sensory Psychology 791

The variety of definitions of stimulus variables 791

The variety of definitions of response variables 797

The range of application of quantitative techniques 798

Variations in the nature and source of concepts 798

References 801

INTRODUCTION

Some difficulties in the statement of methodological problems.

Much conventional methodological discussion in psychology seems to
present gross dissections to the experimenter; not gross in a logical sense,
but gross in terms of providing categories into which the working ma-
terial of the science fits. This feeling probably arises because many of the
analyses of concepts, definitions, theories, etc., have been based on rela-
tively limited samples of the material available for examination, and
those that are selected are far from a random sample. This restrictive-
ness has shown itself, first, in a tendency to pick material that permits a
high degree of compactness and formal simplicity. While such analyses
are instructive when they are first presented, it must be remembered that
the conceptual and theoretical material that forms a simple picture may
not be the material that contains the interesting trouble spots, and if
methodological discussion is to aid in the assault upon scientific questions
rather than serve the functions of a "mopping-up" operation, it will
have to continue to represent the diversities and subtleties of the working

789

790 CONRAD G. MUELLER

material of the science. In this sense, concepts such as force, mass, electri-
cal resistance, the electron, etc., have clearly been overworked. At this
stage of the science it would be much more instructive to encounter a
discussion that could do justice to the plethora of particles in modern
atomic theory of the last decade or two. It might be more helpful to dis-
cuss concepts in psychology, not just with the methodological tools de-
veloped in the context of the classical physics, but with tools able to
handle and illuminate the subtleties of the introduction of the concept
of the meson and the ensuing period of discovery of four or five kinds
of mesons or the invention of a particle as elusive as the neutrino, which
is introduced to save the conservation assumptions and is given a mass of
near zero and a cross section that would permit it to go through our
sun without being detected.

This restrictiveness shows itself in a second way in the stereotypy or
rigidity in the interpretation of what these sample analyses show. For
example, scientists have frequently made excellent attempts at develop-
ing a language for talking about what they do and then acted as if it
were the final language for such a discourse. Perhaps Bergmann [1] has
described a more general trait when he characterizes psychology's re-
sponse to operationism by saying, "The root of the trouble was that
some psychologists in their enthusiasm mistook the operationist footnote
for the whole philosophy of science, if not for the whole of philosophy."
This tendency toward stereotypy is also exhibited in the targets for the
present criticism. In this sense, some of the psychoanalytic concepts and
concepts such as resistance at the synapse have been criticized, if not too
harshly, certainly too frequently. This has created an atmosphere that,
to take the example of synaptic resistance, makes it unlikely that a kind
of literature will be read that has, in the last decade, provided some
evidence that, in fact, there are some relatively long-term changes in the
passage through synapses that result from use. In a variety of ways psy-
chologists have gone along with, and contributed to, this stereotypy with
respect to methodological issues. It is against this background that a dis-
cussion of some of the general characteristics of the sensory area may be
useful.

Some contributions of sensory psychology to methodological problems.
In many ways the sensory area is in a special position with regard to
many methodological questions in psychology. Exactly why this is the
case is not easy to determine. It may be due partly to the fact that the
modern history of this area covers a longer period than is true of most
other areas of psychology. It may also be related in part to the fact that
this history has been more intimately. tied to the history of the physical
and biological sciences than most other areas. Whatever the contribut-
ing factors, there seems to exist a greater diversity of empirical and theo-

Significance of Sensory Psychology 791

retical procedures than is typical of most areas of psychology. The more
closely one examines the sensory area, the more convinced one becomes
that there is little justification for being dogmatic with respect to many
of the issues that form the core of methodological discussions in psy-
chology. For example, one can find ample evidence to argue for an in-
timate link between physiological and behavioral data; there are many
instances in which the physiological data suggested specific experiments
and where physiological theories made specific predictions about be-
havioral data. But one can also find examples of comparable success in
organizing data in the absence of adequate physiological data and in the
absence of such theorizing. There are many examples of useful concepts
derived directly from physical or physiological data or theory, and there
are many concepts that have no such linkage. The same might be said
for other topics, such as operational definitions, etc.

I should like to select four points from the outline that was offered
as a common starting point for the discussions in this study and let these
serve as a focus for considering some general characteristics of the sen-
sory area. These points deal with the questions of definition, quantifica-
tion, and the types and sources of concepts. The topics will be discussed
with the aim of illustrating the diversity of conceptual material that sen-
sory psychology offers for methodological discussion in psychology, in
this way providing a broader informational base upon which methodo-
logical principles in psychology may be formulated.

SOME CHARACTERISTICS OF SENSORY PSYCHOLOGY

The variety of definitions of stimulus variables. There are many
ways of suggesting the full measure of this diversity of conceptual ma-
terial in sensory psychology. An examination of the terms used in the
essays (especially those in Vol. 1) which consider sensory or perceptual
matters would reveal, for example, that stimulus terms can range in
specificity from the highly specific terms encountered when a physical
dimension has been adopted directly, such as distance or energy, to terms
representing broad classes of complex stimuli, such as those referred to
by the terms graininess, timbre, etc. They can also vary in their com-
plexity, i.e., the length of the chain of definitions linking them to experi-
mental procedure, and in many other ways. What is presented below is
a small sample of some obvious differences that may begin to suggest the
scope of the problem.

1. Perhaps the greatest diversity in the description of the stimulus
relates to the extent to which the experimenter adopts the language of
physics. At one extreme the stimulus language does not go beyond the
common extrascientific language. This language is essentially one of ob-

792

CONRAD G. MUELLER

jects — chairs, walls, light bulbs, mountains, railroad tracks — which rep-
resent the descriptive detail in the protocol. Thus, the definition reduces
rather directly to a denotative operation. At the other extreme is the
language of physical dimensions, such as changes in energy, wavelength,
exposure time, etc. The former has led to an emphasis on such experi-
ments as size constancy and the illusions, the latter to a study of the
specific sensitivities, visibility and audibility curves, frequency (audition)
and wavelength (vision) discrimination, etc. Most of the subject matter
in what is loosely called sensory psychology and some of what is called
perception utilize some detailed physical analysis, and again the amount
of analysis is a source of diversity.

2. The stimulus specifications using some physical analysis can differ
with respect to the complexity of their reduction to primitive physical
terms. These terms may involve, as mentioned above, a direct adoption
of primitive physical dimensions, such as distance or time, or they may
involve extensions of such dimensions. As these extensions proceed be-
yond the primitive physical terms, they usually involve more and more of
a commitment to theory, either physical, physiological, or behavioral or
some combination thereof. As an example of the complexity and refine-
ments of some of these terms, consider examples from audition and
vision.

As a first example, consider the concept of frequency as it appears in
physics and in the psychophysics of audition. At the early stages of
measuring something like the audibility curve, a stimulus definition in
terms of elementary physics, e.g., with an understanding of the steady-
state output of an oscillator and its amplitude and frequency, would get
us through most situations. Many of the subtleties of terms like oscil-
lator and frequency can be ignored. But there are many occasions in
audition where the descriptive language becomes more elaborate, not
merely because the physical theory is available for use, but because the
behavioral data suggest that a refinement in stimulus definition or some
comparable adjustment is needed. For example, beginning with an ele-
mentary level of description, it would seem operationally straightfor-
ward to study the frequency sensitivity of the ear for various stimulus
durations. Consider an experiment, for example, that proposes to meas-
ure thresholds at various frequencies ranging from 100 to 1,000 cps and
stimulus durations from 1 msec to 1 sec. Operationally, what one would
probably have in mind is an experiment involving the output of a sine-
wave generator and a mechanism for switching in the sine- wave stimuli
for various lengths of time. With this line of thought, what one would
mean by the experimental proposal stated above is that the stimulus
would have *a waveform with a value of zero up to time to ; up to t\ the
stimulus would be a segment of a sinusoidal function whose period was

Significance of Sensory Psychology 793

specified; after /i the waveform would again return to zero. Even in-
tuitively this problem would seem to get troublesome when one uses
exposure times of the order of 1 msec for frequencies, say, below 500
cps, for here one is presenting a subject with a waveform that is a frac-
tion of 1 cycle. Such a curve seems to lack the obvious properties as-
sociated with the term frequency. More important than one's intuitive
uneasiness is the fact that the discrimination data that he obtains from
this experiment emphasize that there is a problem for the psychologist;
something seems to happen to those psychophysical functions involving
low frequencies and short exposure times [2]. Unfortunately, intuition
does not carry one very far in solving the problem. At this point, mathe-
matical and physical theory may be brought to bear in furnishing some
supplementary language for analyzing the stimulus. This is done, first,
by making the notion of frequency very specific and then, with no loss of
rigor, converting the concept of frequency into one of great generality.
The specificity is achieved by passing beyond intuitive notions of repeti-
tiveness and defining frequency in the following way: any function for
which some nonzero value T can be found, so that f(t) =f(t-{-T} is
true for all values of t, is called periodic, and T is called the period.
Frequency is then defined as 1/71. The generality is achieved by the im-
portant development of Fourier, who showed that any waveform (with
a few restrictions of little relevance to any psychophysical discussion) can
be represented as a set of sinusoidal waves and by showing what the
frequency spectrum of a function /(/) is.

The development may increase the complexity of one's description of
auditory stimuli, but it provides a language of great generality. Regard-
less of how the development is conceived, two considerations must be of
some concern. First, one must employ some device for talking about a
variety of waveforms and waveforms of all durations. Second, the ex-
perimental data require some treatment of the differences among fre-
quencies in their dependence on duration. The question of the usefulness
of this particular refinement in stimulus definition depends on the extent
to which the auditory system is conceived as a frequency-rendering sys-
tem. Perhaps some other type of refinement is required, Licklider, in
Vol. 1 of this series, has presented an interesting and detailed discussion
of the general problem and certain kinds of additional analyses of the
temporal aspects of the auditory system. Such analyses provide the ma-
chinery by which a treatment of auditory discrimination proceeds. No
claim need be made that these analyses add more information about
the stimulus than is contained in a denotative description of the
stimulus waveform; whether this is true depends on the definition of in-
formation. But these analyses do provide specific ways of expressing such
information in terms that will be relevant to subsequent theory.

794 CONRAD G. MUELLER

As a second example., consider a similar type of situation en-
countered in vision in the use of small linear extents. Once again, it
would seem operationally meaningful to investigate the absolute thresh-
old for long thin lines as a function of, say, the width of the line. For
wide lines, an experimental operation such as reducing the width of line
can be interpreted as decreasing the extent of the image formed by an
optical system, such as the eye. But as the width of the line approaches
the dimensions of the order of 1 min of arc such an operation (de-
creasing width) ceases to affect the image size to any appreciable extent,
but rather changes primarily the intensity over a fixed linear extent. This
phenomenon, a problem in diffraction, is common to all optical systems,
and its analysis is directly available in the physicist's treatment of the
nature of light. But once again, the analysis is not applied in the psy-
chophysical situation just because it is available in physical theory, but
because there are psychophysical data that pose certain problems, for
example, the substitutability of area and intensity at threshold, the
general indiscriminability of area and intensity changes for very small
areas, and a variety of problems in visual acuity.

3. The analytical steps in defining the stimulus can also differ in
terms of the extent to which the organism is involved in the definition.
Some terms, regardless of their complexity, can be defined without refer-
ence to the detecting organism. Such would be the case, for example,
with the term frequency discussed above and such terms as wavelength
composition in vision. The raison d'etre for a particular analysis may
involve notions about the organism's behavior or about how some com-
ponent of the organism works, and thus the evolution of the definition
may involve the organism in an important way; but the specification can
proceed without such a reference. Again the concept of frequency illus-
trates this point. One may be interested in a frequency specification of
the stimulus because of a view of how the ear or the auditory system an-
alyzes complex input waves or because of what is known about certain
kinds of auditory discrimination. Nevertheless, one can specify the fre-
quency components of a given waveform without referring to the ear
or the auditory system or its function. A similar situation exists in vision
in the use of angular dimensions. If one measures the absolute threshold
for circular stimuli of varying sizes placed in a dark field, he can obtain
different measurements for each distance the stimuli are placed from
the observer. If, however, one takes some function of the distance and
the size of the stimuli, which turns out to be the angle subtended by the
stimuli, these many functions are unified. Viewed historically, the grad-
ual emergence of this specification was due to a complex interplay of the
data on visual discrimination, the general development of optical theory,
and the physical analysis of how the eye works. This does not alter the

Significance of Sensory Psychology 795

fact that the specification of a term such as visual angle involves no more
of a commitment concerning the organism than where the eye will re-
side in space.

Although many of the definitions of terms encountered in the sen-
sory area can be written without reference to the detecting organism,
there are a large number of remaining terms whose definition requires
some information about how the organism works. This may be seen,
first, in some of the terms that represent what might be called the more
proximal stimulus variables. This increasing involvement shows itself if
one turns his attention from the term visual angle to a term such as the
retinal image, for here it becomes obvious that one is shifting to an anal-
ysis based on notions from physical optics and applying it to the eye
as a physical system. This requires definitive experimental information
concerning the physical properties of the eye, for example, its focal
length. The retinal image cannot be computed without a specific num-
ber refering to this quantity.

This reference to the organism may involve either behavioral data
or physiological data or both, and in the course of the history of the
term it may shift from one to the other. The notion of critical bands in
the frequency tuning of the auditory system is of this sort. One may
evaluate this tuning on the basis of psychophysical data, such as masking
data or data on frequency discrimination, or he may, after the ingenious
experiments of von Bekesy [6] on the mechanical properties of the inner
ear, directly evaluate this tuning on the basis of physical measurements
of basilar membrane vibrations.

The involvement of the organism in such dimensional definitions can
become very complex. One of the most complex examples in the sensory
area is the specification of color stimuli in vision; the reader is referred
to Graham's article (Vol. 1) for this intricate example. Another, and
perhaps more familiar, example in vision is the class of dimensions called
the photometric dimensions. These dimensions are described by Judd
[5] in the following way :

If it is desired to convert the radiant flux (watts) entering the pupil
of the eye to luminous flux (lumens), the additivity law is applied. The
radiant flux is analyzed spectrally so that for each portion A A of the spec-
trum the spectral radiant flux PA is known. Then, by multiplying the
spectral radiant flux by the absolute luminosity KX (lumens per watt) for
that wavelength region, we find the spectral distribution of luminous flux.
But, by the additivity law, the total luminous flux F is equal to the sum of
the parts making up this spectral distribution, thus :

00

F = y PX.KXAX

796 CONRAD G. MUELLER

where AX is a wavelength interval so small that further reduction fails to
alter the sum significantly.

With respect to this so-called additivity law and the general problem of
the psychophysical discriminations involved in this type of dimension
Judd also says:

The conversion of a stimulus specification from radiant to luminous
terms is based upon the additivity "law55 of luminance: if a stimulus of
luminance J5t (such as is produced by a spot of light on a screen) is added
to a second stimulus of luminance B2 (such as is produced by a second
spotlight shining on the same screen), the luminance B of the combination
stimulus is defined as the sum of the luminances of the component stimuli;
that is, B = #! + #2- This law has frequently been studied because it is
the basis of photometry (Dresler, 1937; Kohlrausch, 1935; Urbanek and
Ferencz, 1942). It has been found to fail unless the eye is kept throughout
the series of comparisons essentially in a fixed state of adaptation. That is
to say, the law fails unless there is pure cone vision, pure rod vision, or
some constant combination of the two. Some reports indicate that it fails
anyhow.

If a spot of red light is adjusted to the same brightness as a spot of
yellow light, and a spot of green light is similarly adjusted to match the
brightness of a second yellow light, the red and green lights added together
are often found to be darker than the sum of the two yellow lights (Dresler,
1937). Since it is impossible for an observer to report with high precision
and reproducibility which of two spots of light of widely different chromatic
character is the brighter, these failures of the law have not been taken very
seriously. The usual explanation is that the observer mistook the high satu-
ration of the red field, relative to that of the yellow, for brightness and so
obtained a spuriously high estimate of its luminance in the first place. The
next time the observer tests the additivity law his observations are somewhat
conditioned by the first experience, and soon he has learned to make photo-
metric settings in accordance with the law. Thus the additivity law provides
a basis for a convenient photometric technique that correlates excellently,
though not perfectly, with visual experience.

Thus, the status of this kind of variable involves the interpretation of
mistakes the subject may make (confusion of saturation and brightness)
and the gradual acquisition of the behavior that is in accordance with
the "law" utilized in the definition.

In a sense, a photometric dimension is a "physical" dimension, in the
trivial sense that probably all dimensions in a behavioral science are
"physical" It is an energy function in wavelength coupled with a
"weighting" function. The weighting function is a convention — it is the
"agreed-upon" variation of sensitivity with wavelength for the "average"
observer. Thus, it is observer-determined, although it is not a correction
on an individual basis. These photometric dimensions have many ad-

Significance of Sensory Psychology 797

vantages in that they unify many disparate functions when wavelength
is a parameter. They also have some disadvantages, particularly when
their origin and definition is forgotten. For example, if one plots a series
of dark-adaptation curves for various wavelengths and uses a typical
photometric unit, he observes that the curve showing the greatest drop,
and the lowest terminal threshold, is one representing the blue end of
the spectrum. In the past, this has led to the statement that the dark-
adapted eye is the most sensitive in the blue end of the spectrum. Since
sensitivity is quite generally defined as the reciprocal of the energy re-
quired at threshold, this statement is clearly incorrect, as any of the
determinations of the dim-visibility curve will show. What happens in
such a treatment of the data is that the photometric units used involve
a correction based on the visibility curve for the light-adapted eye. Un-
fortunately, the dark-adapted eye exhibits a different visibility function,
and what such a graph actually shows is a combination of what might
be called the basic change in sensitivity, defined as indicated above, and
the change from one visibility function to another. The differences among
the different wavelengths reflect the magnitude of this difference in the
"weighting" term at different stages of adaptation.

The variety of definitions of response variables. The diversity of
description of the response terms in experiments in sensory psychology is
more difficult to evaluate because less is known about what the problems
are and how to treat the subject matter. There are differences, first of
all, in the manner in which response-class membership is determined.
Some experiments involve an apparatus component for defining the
topography of the response being measured. For example, the subject
may be asked to press one of n keys or turn one of n knobs. In other ex-
periments the experimenter performs this function. For example, the
subject may be asked to give one of n verbal responses, say, yes-no,
large-small, very heavy-heavy-medium-light-very light, or he is asked to
assign a number from one to ten or one to one hundred to the stimulus.
Here the experimenter plays the key role identifying instances of the
class of responses used. From the point of view of a detailed analysis of
the sensory experiment much needs to be done to clarify the relation of
these two procedures to each other and to the related studies in lower
animals.

Even more complex in terms of any theoretical analysis are those ex-
periments in which the experimenter reports what he sees or hears when
he performs some operations. While it is true, as pointed out by Licklider
(Vol. 1), that this kind of observation will continue to act as an im-
portant guide for the experimenter's behavior, there can be little doubt
that such a procedure offers difficulties for any attempt to outline a
rigorous behavioral account of this experimental area.

798 CONRAD G. MUELLER

The range of application of quantitative techniques. The sensory
area is also diversified with respect to the quantitative techniques used,
and this diversification is one of degree and kind. The degree of quanti-
fication can range from its complete absence to some of the highly for-
malized techniques illustrated in Vol. 1. Some examples are to be found
in the articles by Blank, Graham, Licklider, and Pirenne and Marriott.
As examples of this diversity one encounters elements of matrices and
determinants in the development of the color equations for specifying
color stimuli; one sees the notion of the line integral in certain theories
of color vision; there are many uses of ordinary differential equations in
the many versions of photochemical theory in vision as exploited by
Hecht [4] and extended by many others. One sees the utilization of many
aspects of probability theory, e.g., the theory of random time processes,
as discussed extensively by Pirenne and Marriott, and the use of certain
specialized probability notions, such as information theory and decision
theory, in the theories of the ideal observer and certain formulations in
perception. One encounters the use of the Fourier integral in many
problems in audition, e.g., frequency analysis as discussed above and in
the article by Licklider; one sees also some of the general features of
measure theory operating in establishing geometrical properties of a
hypothetical visual space so clearly illustrated in the article by Blank.

These quantitative procedures serve a variety of functions. They may
represent refinements in the descriptive language of the stimulus, they
may provide a specialized technique for discussing stimulus-response cor-
relations, or they may offer a means of solving for functional relations
between theoretical variables.

Variations in the nature and source of concepts. Finally, it is prob-
ably with respect to the nature and source of concepts that sensory psy-
chology can provide its most unique emphasis in a methodological dis-
cussion. While there are many concepts that are primarily behaviorally
derived and that perform functions similar to those discussed under the
heading of intervening variables in the MacCorquodale and Meehl
sense, there are numerous specific concepts, with many quantitative em-
pirical and theoretical properties, that owe their existence and properties
to data and experiment in another discipline. There are many examples
of such concepts in the fields of audition and vision. For example, a
theory by Helmholtz suggested that the ear acts as a series of tuned
resonators. The phrasing of the theoretical statements and the context in
which they were introduced carried the implication that the basilar
membrane would possess certain physical properties that would permit
this resonance, e.g., that the basilar membrane would exist under greater
lateral than longitudinal tension. Following the formulation of Helm-
holtz there were many different conceptions of how the ear worked. In

Significance of Sensory Psychology 799

addition to the helmholtzian resonance model, there was a class of
theories that conceived of the cochlear canals as tubes filled with fluid
with the basilar membrane serving as an elastic partition. Many differ-
ent theories resulted within this framework, depending on the physical
properties assumed for different parts of the inner ear. If one assumes a
certain damping, he can generate a series of standing waves, i.e., a
different standing-wave pattern for each frequency. With other as-
sumptions, traveling wave patterns can be generated that have maxima
at different positions for different frequencies. If the elastic partition is
considered to be relatively stiff, the whole length will vibrate approxi-
mately in phase, and one has a vibration mechanism something like a
microphone. All these assumptions can lead to statements about the
discriminability of pitch and other psychophysical data, but it is obvious
that they can survive simultaneously only in a factual vacuum regard-
ing the physical properties of the ear. With the advent of the now classic
experiments of von Bekesy on the mechanical properties of the inner ear,
this freedom of conceptualization is restricted. The physical consequences
of a theory such as that of Helmholtz are not observed; von Bekesy's re-
sults suggest that the basilar membrane is not under significant tension.
The consequences of a telephone or microphone conception of the
basilar vibration are also not born out; the vibration of all parts of the
membrane are not in phase. Thus, the ways of generating predictions or
"explanations'3 of the psychophysical data that have a common starting
point in viewing the ear as a physical and physiological system can be
separated on the basis of nonpsychophysical experiments such as those
of von Bekesy. At the same time these data can provide the basis for
further extensions of the theory or theories that survive the fact gather-
ing. Licklider's article clearly indicates this interplay of the behavioral
and physiological data in the molding of auditory theory and experi-
ment.

The notions of photochemistry behind many theories in vision also
emphasize this complex interplay between psychophysical data and
theory on the one hand and physiological data and theory on the other.
The concept of a photochemical action in vision arose early in the
modern history of photochemistry itself. The visual aspects of this prob-
lem date back to the discovery of visual purple by Boll in 1876, although
it was really Kuehne (1879) who gave the details of visual-purple ex-
traction and studied many of its physical properties. It became clear very
early that this material in the retina was bleached in the presence of
light and regenerated in the dark. This seemed to offer an interesting
link with what was known about the changes in sensitivity of the human
eye in light and dark adaptation — light, in bleaching visual purple, left
less photosensitive material available to absorb light. Thus, to get the

800 CONRAD G. MUELLER

same photochemical effect before and after light adaptation, it was
necessary to present more light after light adaptation. A second line of
evidence was also of great significance. Koenig, as early as 1894, first
clearly showed that a quantitative agreement existed between the visi-
bility data (i.e., the sensitivity of the eye to different wavelengths) and
the absorption curve of visual purple. A decade later Trendelenberg
showed that the rate of bleaching of visual purple also agreed with the
visibility curve; that is, regions of the spectrum easily seen are regions
that bleach visual purple rapidly. This kind of correspondence led to a
number of quantitative formulations of the action of the visual system
based on the kinetics of a regenerative photochemical system.

The important point is not that such correspondences (and there are
many others), as suggested in the preceding paragraph, exist but rather
that they are part of a joint, dependent development of two sections of
science, behavioral studies of what one sees and the physiological and
biochemical studies of the visual system. The behavioral data influenced
physiological research and the physiological data influenced the be-
havioral research. For example, one form of this photochemical con-
ceptualization as formulated by Hecht in the early 1920s conceived of a
photochemical-chemical cycle wherein light bleached a photosensitive
substance, and this bleached material, in the presence of other materials,
reformed the photosensitive material via a chemical (nonphotic) re-
action. Since the regenerative link in this cycle was a purely chemical
one, it was expected to exhibit certain properties characteristic of chemi-
cal reactions, and this suggested a number of specific directions in which
research might proceed. Since the velocity of reactions of the type en-
visioned are typically temperature dependent and since the regenerative
link was viewed as critical in the increasing sensitivity as a function of
time in the dark (the psychophysical problem of dark adaptation),
Hecht [3] predicted that dark-adaptation curves would be more rapidly
changing functions at high temperatures than at low temperatures. This
problem was studied behaviorally in cold-blooded animals; the outcome
of the experiments was that the dark-adaptation curves were tempera-
ture dependent in the predicted way.

A second, and more striking, example in vision centers on a large
number of human behavioral experiments dealing with the so-called
Wald and Clark effect. By the middle 1930s the biochemical research
had shown that the visual-purple cycle was more complicated than
originally conceived and that it possessed at least three important stages:
visual purple, retinene, and vitamin A (and perhaps a number of addi-
tional transitional stages). The data clearly indicated that the velocities
of the changes from one stage to another were not identical and that it
was possible to vary the procedures for presenting the incident light in

Significance of Sensory Psychology 801

such a way as to yield different combinations of concentrations of the
various substances involved. For example, the data indicated that it
should be possible to find two adaptation procedures that would produce
the same resultant concentration of visual purple but yield different com-
binations of concentrations of retinene and vitamin A. The data on the
velocity constants for the component reactions clearly suggested that the
rates of regeneration of visual purple in these two cases would be differ-
ent; the rate of return to visual purple from vitamin A was different
from the rate of return from retinene. Coupled with the general assump-
tion of the photochemical theory of that period that sensitivity was a
function of the concentration of the photosensitive material, this bio-
chemical research led to an important set of expectations with respect
to dark-adaptation curves. The general expectation was that adaptation
curves that started at the same point could be made to follow a variety
of courses to terminal threshold, depending on the prior conditions of
adaptation. The Wald and Clark experiment and a large number of
experiments that followed have confirmed and elaborated this point [7].
Needless to say, the preceding paragraphs should not be interpreted
as saying that concepts arrived at in this manner, such as the concepts of
a photochemical theory, etc., are the correct concepts; they may be
right or wrong in the same sense and in the same ways as concepts
derived from other behavioral data. Rather, two points may be empha-
sized. The first is that data and theory of a physiological sort can and do
guide and generally interact with behavioral experiments and programs
of research in the sensory area in the same way that one set of behavioral
data or concepts will interact with another. No new questions about
"levels of discourse" need enter into this interaction that are not en-
countered in many other contexts in psychology where one moves from
one "level" to another, for example, the interpretation of the complex
topography of many human problem-solving situations in terms of a
simpler and more easily specifiable topography of an animal acquisition
experiment. The second point is a return to the emphasis on diversity,
for there are many examples where this interaction with physiological
data has been a minor factor. For example, most of the developments
that have taken place in color vision have been little supported by spe-
cific physiological data.

REFERENCES

1. Bergmann,, G. Sense and nonsense in operationism. Sci. mon., 1954,
79,210-214.

2. Garner, W. R. The effect of frequency spectrum on temporal integra-
tion of energy in the ear. /. acoust. Soc. Amer., 1947, 19, 808-815,

802 CONRAD G. MUELLER

3. Hecht, S. The kinetics of dark adaptation. /. gen. Physiol, 1927, 10,
781-809.

4. Hecht, S. Rods, cones and the chemical basis of vision. Physiol. Rev.,
1937, 17, 239-290.

5. Judd, D. B. Basic correlates of the visual stimulus. In S. S. Stevens
(Ed.), Handbook of experimental psychology. New York: Wiley, 1951.

6. Von Bekesy, G., & Rosenblith, W. A. The mechanical properties of
the ear. In S. S. Stevens (Ed.), Handbook of experimental psychology. New
York: Wiley, 1951.

7. Wald, G., & Clark, A. B. Visual adaptation and the chemistry of the
rods. 7. gen. Physiol., 1937, 21, 93-105.

NAME INDEX

Page numbers in boldface type indicate bibliography references; <en." indicates foot-
note reference.

Abraham, K., 121n., 157, 167

Adams, D. K., 53, 132, 167

Adler, A., 20, 53, 57, 79, 102, 133

Adorno, T. W., 472

Ahmavaara, Y., 279, 315, 320, 322

Ainsworth, M., 175

Aitken, A. G., 542

Alexander, F., 13, 53, 133, 138, 155, 167,

171, 615n., 709
Allen, G., 335, 336, 337, 352, 355, 357,

358, 360
Allport, F. H., 369-370, 383, 409, 420,

427, 467, 472, 473, 709
Allport, G. W., 28n., 53, 74, 85n., 88,

96, 97, 115, 122, 126, 145, 167,

322, 389rc., 420, 434, 439, 472, 481,

542, 621, 622
Alper, T. G., 28n.
Alstrom, C. H., 353
Alter, J., 479
Anastasi, A., 358
Anderson, E. E., 293, 322
Anderson, G. H., 322
Anderson, H. H., 322
Anderson, J. C., 283, 323
Anderson, R., 219, 252
Anderson, T. W,, Jr., 542
Andrews, T. G., 283, 322
Angyal, A., 196, 198, 252
Arensberg, C. M., 548
Aristotle, 12, 29, 37, 49, 53, 330
Asch, M. J., 242, 252, 256
Asch, S. E,, 5, 363, 383, 443, 472, 750,

762, 768, 771, 780
Ashley, W. R., 459, 472
Assum, A. L., 219, 252
Axline, V. M., 243, 252

Bachrach, A,, 169

Back, K. W., 410, 415, 420, 472

Baggaley, A. R., 267, 279, 302, 316, 323

Bain, R,, 427, 472

Bakcn, D., 62n.

Bales, R. F., 28n., 548, 596, 608, 609,

623, 624, 678ra., 690, 693, 707n.,

710

Bargmann, R., 279, 281, 286, 316, 322
Barker, R. G., 549

Barlow, M, F., 420

Baroff, G. S., 346, 358, 359

Barrett, W. G., 28n.

Barton, A. H., 542

Bauer, J., 339, 358

Beach, F. A., 69, 167

Beadle, G. W., 358

Becker, G. L., 338

Bell, R. O., 286, 322

Bellak, L., 28n.

Beloff, J. R., 315

Ben-Zeev, S., 545n., 580, 606, 608, 611

Benjamin, J. D., 66n., 69n., 106, 140,

141, 158, 167

Berelson, B., 421, 543, 709rc.
Bergman, P., 143n., 167
Bergmann, G., 54, 790, 801
Bergson, H., 7, 24, 38, 53, 290
Bernfeld, S., 137, 151, 162, 167, 168
Bernfeld, S. G., 168
Beraheim, H., 60, 61, 112, 168
Bettelheim, B,, 104, 157, 168, 449, 472
Beveridge, W. L B., 165, 168
Bexton, W. H., 168, 175
Bibring, E., 97n., 15 In., 156, 168
Bigelow, N. H., 182
Binger, G., 138
Bion, W. R., 545, 548-549, 555, 573,

574, 577-578, 580, 581, 583-584,

590, 595, 596, 605, 609
Blake, R., 139, 168
Blank, A. A., 772, 798
Blewett, D. B., 315
Blocksma, D., 597, 609
Bogardus, E. S., 420, 431, 455, 529
Bohr, N., 162
Boll, 799

Bond, D. D., 123n., 168
Borgatta, E. F., 609, 710
Boring, E. G., 19, 139, 168
Bown, O., 194
Brenman, M., 72, 98, 106, 139, 141, 168,

174

Brentano, F., 60, 61
Breuer, J., 60, 61, llOn., 112, 126, 168
Bridgeman, P. W., 54, 709
Brodbeck, A. J., 709
Brodbeck, M., 542
Brogden, H. E., 281, 322

803

804

NAME INDEX

Brown, J. F., 160, 168

Brown, J. S., 300

Brownfain, J. J., 477, 472

Bruner, J. S., 108, 131, 139, 168, 169,

429, 434, 459, 475
Brunswik, E., 27, 54, lOln., 108, 122,

122n., 166n., 738
Buehler, K., 96, 169
Burdick, E., 709
Burks, B. S., 344, 345, 358
Burlingharn, D., 339, 340, 358
Burt, G., 259, 260, 322
Butler, J. M., 194, 208, 219, 247, 252
Butler, S., 6 In.

Gannon, W. B., 17-18, 723 625, 710

Gantril, H., 301

Carlson, E. R., 461, 472

Carnap, R.3 484, 485, 486, 487n., 542,

747n., 788

Garr, A. C., 219, 247, 252
Cartwright, D., 194, 219, 232, 244n.,

252, 253, 386, 420, 432, 436, 443,

472, 547, 609, 745, 747, 750, 751,

762, 768, 773, 780
Cassirer, E., 45, 53
Gattell, A. K. S., 323
Gattell, J., 54
Gattell, R. B., 5, 257, 267, 279, 283, 297,

302, 315, 316, 319, 322, 323, 324,

332, 337, 358, 547-548, 609, 738,

745, 772, 775
Chance, J., 219, 252
Chappie, E. D., 548
Charcot, J. M., 60, 61, 62, 112, 112n.
Charters, W. W., Jr., 414, 420, 472
Ghave, E. J., 475, 534, 543
Chein, I., 86n., 169
Chodorkoff, B., 202, 217, 232-233, 245,

252

Chowdhry, K., 420
Christie, R., 147n.
Ghrobak, R., 61
Cicero, 330

Clark, A. B,, 800, 801,802
Clark, K. B.3 447, 472
Clark, K. E., vi, via
Clark, P. J., 334
Coan, R., 315
Coch, L.3 242, 252
Gofer, C.N., 219, 252
Cohen, A. R., 438, 472
Collier, R. M., 463, 472
Collins, M. E., 410, 420
Combs, A. W., 194, 197, 219, 253
Conant, J. B., 625, 626, 709, 710
Conrad, G., 353, 358
Coombs, C. H., 269, 285, 306, 324
Copernicus, N., 63
Cowen, E. L., 219, 253
Criswell, J., 549
Crockett, W. H., 460, 472
Cronbach, L. J., 289, 324, 358

Cross, P., 323

Crutcher, R., 281, 324

Crutchfield, R. S., 389rc., 421, 427, 428,

430, 443, 474

Culbertson, F. M., 458, 463, 473
Culler, E., 171
Cummins, H., 358
Cureton, E. E., 270, 324
Curie, M., 190
Curie, P., 190

Dahlberg, G., 358

Damarin, F. L., 312

Darlington, G. D., 358

Darwin, G., 7, 14, 20, 38-41, 45, 46, 53,

60, 63, 68, 69, 73, 78, 80ra.
Das, R., 297, 300, 324
David, H. P., 358
Davids, A., 28n.
Day, M., 289
Decher, H., 362
de Haan, R., 545n., 605, 609
Delbrueck, M., 162, 169
Dembo, T., 28n.
de Saussure, R., 137, 182
Deutsch, F., 138, 155, 169
Deutsch, M., 410, 420, 549, 609
Dewey, J., 186, 208, 480, 481, 542,

572n., 778
Dickman, K., 316
Dickson, W. J., 242, 255, 422
Diller, L., 232, 253
Dingle, H., 165, 169
Diogenes, 330
Diven, K., 28n., 68, 169
Doane, B. K., 175
Dobzhansky, T., 358
Dollard, J., 45, 54, 79, 86, 137, 140,

144, 145, 146, 158, 169, 178, 383
Doob, L. W., 383
Downey, J., 281
Dresler, A., 796
Driver, R., 312
Dubin, S. S., 282, 323
DuBois, C., 28n.
Dunbar, H. F., 138, 155, 169
Dunn, L. C., 358

Durkheim, E., 467, 468, 620, 656, 691
Dyk, W, 28n.
Dymond, R. F., 219, 248, 252, 253,

255

Edward, A. L., 473

Einstein, A., 63, 8472., 135, 698

Eissler, K. R., 156, 169, 176

Ekman, G., 324

Ellis, W. D., 131, 169

Ellison, D. G., 270, 288, 290, 725, 772,

773, 774, 781, 782
Emerson, A. E., 50
Empedocles, 330
English, O. S., 155, 183

NAME INDEX

805

Erickson, E. H., 28n., 53, 57, 59, 60n.,
62, 63, 67, 69, 79, 85, 96, 97n.9 100,
103, 103n., 104, 122n., 134, 143,
150, 1523 152n., 154, 156, 158, 159,
160, 169, 170, 691

Erkkson, M. H., 67, 139, 169, 176

Escalona, S., 140, 141, 158, 170

Essen-Moller, E., 345, 359

Estes, W. K., 383, 542, 743, 751, 758,
772, 774, 782

Euclid, 135, 779

Eysenck, H. J., 262, 282, 283, 285, 314,
315, 324, 351, 359

Ezriel, H., 548

Farber, I. E., 300

Farber, L., 139, 170

Faw, V., 242, 253

Fechner, G. T., 60n., 170, 309

Federn, P., 157, 170

Feigl, H., 19, 54, 119, 170, 542, 788

Feingold, L. A., 346, 359

Feldstein, M. J., 411,422

Feller, W., 542

Fenichel, O., 87n., 92rz., 102n., 103n.,
134, 157, 170

Ferencz, E., 796

Ferenczi, S., 99, 170, 171

Ferguson, G. A., 315, 324

Feshbach, S., 460, 473

Festinger, L., 386, 407, 410, 415, 420,
447, 473

Fiedler, F. E., 215, 253, 285, 324

Finch, G., 171

Fisher, G., 68, 139, 142, 170, 171

Fisher, R. A., 359

Fiske, D. W., 262, 324, 325

Flanagan, J. G., 284n., 324

Flanders, N. A., 546-547, 597, 605, 609

Fleichman, E. A., 313

Fliess, W., 80n.

Foote, N., 157

Ford, N., 359

Fourier, J. B. J, 793

Frank, J. D., 28n., 34

Freedman, M., 606, 609

Freeman, F. N., 345

French, J. R. P., Jr., 242, 252

French, J. W., 281, 324

French, T. M., 85, 86, 133, 138, 141,
155, 158, 171

French, V. V., 28rz.

Frenkel-Brunswik, E., 54, 105n., 119,
123, 171, 468, 472

Freud, A., 61n., 100, 103, 134, 156, 171

Freud, S., 7, 13, 20, 27, 29, 36-38, 53,
55, 57, 58, 59, 60, 61, 62, 63, 64,
65, 66, 67, 68, 69, 70, 71, 72, 73,
74, 75, 76, 77, 78, 79, 80, 80n.,
83n., 85, 87n., 89, 92rz., 93, 94, 98,
99, 100, 101, 102, 103, llOn., llln.,
112, 113, 115, 117, lain., 126,
127n., 132, 133, 134, 136, 137, 139,
140n., 141, 145, 150, 152, 155, 156,

157, 163, 168, 171, 172, 173, 191,
248, 260, 467, 468, 548, 621, 622,
624, 653, 654, 655n., 656, 658, 680,
691, 692, 710, 784

Frick, F. C., 772, 773, 774, 781

Fried, G., 2Bn.f 33

Friedman, G. R., 170, 177, 180

Fries, M., 158

Fromm, E., 102n., 103n., 133, 134, 157,
622

Fromm-Reichmann, F., 157, 173

Fruchter, B., 324

Frumkin, S., 359

Fuller, J. L., 359

Galen, 330

Galileo, 28, 29, 135, 625, 701, 779

Gallagher, J.J., 215, 253

Gallon, F., 259, 331, 344, 345, 359

Gardner, J., 28n.

Gardner, R. W., 109, 173

Garner, W. R., 801

Gaudet, H., 421, 543

Gedda, L., 344, 359

Geertz, G., 708

Gendlin, E., 194

Gerard, R. W., 50, 359

Gesell, A., 331, 344, 359

Gibson, J. J., 122, 729, 7503 760-761,

768, 780

Gibson, W., 533-534, 539n.
Gibson, W. A., 270
Gill, M. M., 59, 72, 79, 88, 98, 104, 106,

123, 139, 141, 143n., 144n., 154,

156, 159, 168, 174, 181
Gilliam, S., 542
Ginsberg, M., 620
Gladstone, A. L, 457, 473
Gleser, G. G., 324

Glidewell, J. C., 545n., 605, 606, 609
Glover, E., 78, 174
Goethe, J. W., 59, 62
Goldiamond, I., 232, 253
Goldmann, A. E., 174
Goldschmidt, R. B., 341, 359
Goldstein, K., 85, 85n.
Goodman, G., 108, 131, 168, 459
Gordon, T., 194, 219, 242, 248, 250,

253

Gough, H., 391, 420
Gourevitch, S., 176
Gradolph, P., 545n., 606, 611
Graewe, H., 344, 359
Graham, C. H., vii, 6, 759, 760, 772,

795, 798

Grandolph, L, 545n., 611
Grebe, H., 359
Green, B. F., 539, 542
Greenacre, P., 168, 176, 183
Greisinger, W., 60
Grinker, R. R., 80n.
Gross, A., 60n., 151, 162, 174
Gross, L., 242, 253

806

NAME INDEX

Gruen, W., 282, 315, 323

Grummon, D. L., 219, 247, 253

Guett, A., 358

Guilford, J. P., 277, 280, 292, 314, 324,

325

Gulliksen, H., 770
Guthrie, E. R., 738, 747, 751, 756, 757,

759, 763, 764, 765, 767, 771, 779
Guttmann, L., 319, 430, 529, 532

Haeckel, E. H., 60n., 69

Haigh, G. V., 217, 219, 252, 253

Haimowitz, M. L., 219, 220, 253

Haimowitz, N. R., 219, 253

Hall, G. S., 54

Halverson, H. M., 140, 174

Hanlon, T. E., 219, 254

Harary, F., 386, 420, 443, 472

Harding, J., 458, 473

Hare, A. P., 609, 710

Hare, G. G., 360

Harlow, H. F., 147n., 293, 325, 725, 751,

767

Harper, R. S., 459, 472
Hartley, E. L., 420, 421, 472, 473
Hartley, M., 219, 254
Hartman, D. A., 472
Hartman, G. W., 461, 473, 475
Hartmann, H., 57, 59, 60n., 67, 69, 79,
79n., 83, 92n., 94, 95, 96, 96n.,
977Z., 98n., 100, lOQn., 101, 1033
103n., 104, 107, 122, 134, 137, 147,
150, 152, 152rc., 153, 154, 156, 157,
158, 159, 160, 174, 175
Hartshorne, H., 300, 325
Haverland, E. M., 293, 325
Hebb, D. O., 69, 72, 109, 124n., 131,
132, 175, 199, 650, 751, 767, 769,
771, 772

Hecht, S., 798, 800, 802
Hegel, G. W. F., 63

Heider, F., 80, lOln., 108, 122, 122n.,
375, 383, 386, 407-408, 416, 420,
421, 443, 472, 473
Helmholtz, H. L. F., 60, 63, 80n., 798,

799

Helson, H., 767, 772
Hempel, G. G., 484, 485-486, 487, 542
Hemphill, J. K.5 548
Henderson, A. M., 711
Henderson, L. J., 7, 17, 45, 53, 54, 625,

626, 710
Heraclitus, 23
Herbart, J. F., 61
Hering, E,, 61
Heron, W., 168, 175
Herrman, L., 344, 345, 359
Hilgard, E. R., 140, 142, 175, 202, 254
Hill, W. F., 545n., 606, 609, 611
Hinde, R. A., 750, 751, 767, 771
Hippocrates, 53, 330
Hobbes, T., 372
Hobhouse, L. T., 620

Hoffman, A. E., 220, 254

Hofstaetter, P. R., 219, 254, 315, 325

Hogan, R., 194, 217,219,254

Hogben, L., 344, 345, 359

Hogrefe, R., 458, 473

Hollingworth, L., 186

Holt, E. B., 137 n., 175

Holt, R. R., 28n., 59, 8072., 1247Z., 128,

129, 175, 176
Holzinger, K. J., 337, 345
Holzman, P. S., 109, 175
Homans, G. C., 406-407, 408, 409, 421,

548, 565, 607, 609
Hooker, D., 358, 360
Horn, D., 28ra.
Horney, K., 53, 57, 67, 79, 102, 133,

143n., 175, 622
Horowitz, M. W., 421
Horst, P., 270, 288, 325
Horwitz, M., 592, 609
Hovland, G. I., 421, 462, 473
Hull, G. L, 54, 79, 137, 175, 260, 300,

325, 365, 383, 483, 623, 696, 710,

734, 739, 742, 758, 770, 773
Hume, D., 90
Humphrey, G., 137n., 175
Humphreys, L. G., 296
Hunt, J. McV., 69, 140, 142n., 176
Hunter, W. S., 75
Husen, T., 339, 359
Huston, P. E., 67, 176

Jackson, H., 60, 69, 70, 73, 78

Jacobson, E., 79, 97n., 123n., 176

James, W., 53, 72, 479, 480, 481, 542

Janet, P., 53, 61, 70, 112n.

Janis, I. L., 421, 457, 458, 460, 473

Janowitz, M., 449, 472

Jaques, E., 28n.

Jaynes, J., 69, 167

Jennings, H. H., 549, 609

John, E. S., 219, 247, 253, 254

Jolly, D. H., 360

Jones, E., 176

Jones, M., 176

Jones, M. R., 474

Jonietz, A., 219, 220, 254

Jordan, N,, 407, 416, 421

Judd, D. B., 795, 796, 802

Jung, G., 7, 13, 36, 53, 57, 133, 137, 280

Kahn, R. L., 454, 473

Kaiser, H., 134

Kallmann, F. J., 5, 260, 328, 358, 359,

360, 725
Kant, L, 62, 90

Kaplan, A., 328n., 486, 487, 542
Kapos., A., 421
Kardiner, A., 57, 67, 79, 102, 103n., 133,

152, 154, 176, 622
Karn, M. N., 360

NAME INDEX

807

Katz, D, 5, 260, 423, 454, 463, 473,

7455 762, 771
Kauffman, P. E., 219, 254
Kaufman, R, 138
Keith, A., 46
Kelley, E. C., 254
Kelley, E. L., 262, 325
Kelley, H. H., 414, 420, 421
Kelman, H. C., 441, 462, 473
Kempf, E. J., 137n.9 176, 325
Kendig, I. V., 28n.
Kendler, H. H., vi, vii, 6
Kepler, J., 135, 779
Kessler, C., 217, 254
Kilpatrick, W. H., 186
King, B. T., 458, 473
Kirtner, W. L., 214n., 254
Klapper, Z, 360
Klein, G. S, 59, 68, lOln., 106, 109,

122, 122n., 131, 139, 159, 175, 176
Klein, M., 133, 134
Klisurich, D., 242, 255
Klopfer, B., 175
Klopfer, W., 175
Kluckhohn, C., 45, 53, 157, 170, 622,

710

Kluckhohn, F.5 45, 53
Knight, R. P., 14472, 170, 177, 180
Knower, F. F, 461, 473
Koch, S., vi, vii, 54, 57, 58, 59, 105, 107,

116, 150, 361, 383, 483-484, 542,

729

Koenig, A, 800
Koffka, K., 26, 10-8, 177
Kohler, W., 75, 376, 383, 620n, 623n.,

710, 736, 768
Kohlrausch, A., 796
Koht, A., 28n.
Korzybski, A, 34, 567, 610
Kramer, B. M, 474
Kranz, H., 353, 360
Krech, D., 389n., 421, 427, 430, 443,

474
Kris, E., 53, 61n., 69, 79, 92n., 95, 100,

134, 150, 158, 175, 177
Kroebcr, A. L., 710
Krus, D, 177

Kubie, L. S., 138, 141, 169, 177
Kuehne, 799
Kutner, B., 461, 474

LaBarre, W., 573n., 610
Lamarck, JM 60, 69
Lamy, M., 360
Landauer, W., 360
Lange, G., 72
Lange, J., 344, 353, 360
Langcr, W. C., 28n.
Lanier, L., vi, vii, 6
LaPiere, R. T., 461, 474
Lasswell, H. D., 157
Laswell, H., 45, 53
Lau, J. B, 421

Lauterbach, C. E., 344, 345, 360

Lawrence, D. H., 296

Lazarsfeld, P. F., 5, 421, 476, 542, 543,
709n., 737, 738, 750, 772, 774, 775

Lazarus, R. S., 199, 254, 429, 474

Leary, L, 710

Lecky, P., 202, 254

Lee, M. G., 286, 325

Leeper, R. W., 77n., 177

Leibniz, G. W, 38

Leitch, M., 158

Lennox, W. G., 353, 360

Levine, I.s 6 In.

Levine, J. M., 474

Levinson, D. J., 472

Levy, D. M., 140, 142n., 177, 186

Levy, S. J, 219, 252

Lewin, B., 157, 177

Lewin, K.3 7, 13, 20, 21-29, 34, 53, 54,
75n., 79, 80, 86, 86n., 93, 106,
106n.3 121, 122, 122/2., 125, 131,
158, 161, 177, 178, 421, 549, 591,
593, 599, 610, 745, 747, 750, 768,
770, 773

Lewin, M., 138

Lichtenberg, P, 28n., 33

Licklider, J. G. R., 748, 751, 759, 760,

772, 781, 793, 797, 798, 799
Lieberman, S, 458, 474
Liebermann, M., 545n.9 606, 610
Lilienfeld, A. M., 360

Lilly, J. C., 72, 178

Lindzey, G, 28n., 54

Linton, R.3 45

Lipkin, S., 215, 217, 254

Lippitt, R., 610

Livingston, G., 28?2.

Loewenstein, R. M., 79, 92n, 95, 134,

175, 176
Logan, F. A., 742, 743, 751, 758, 772,

773, 781, 782
Lorand, S., 178

Lord, F., 506n., 541 n., 543

Lorenz, K., 69, 178

Lotka, A. J, 50, 54

Lovell, G. D., 277

Lowrey, L., 186

Lubin, A., 270, 286, 288, 306, 325

Lucretius, 38

Lumsdaine, A. A., 457, 462, 473

Luxenburger, H., 345, 360

Lyons, J., 421

McArthur, G. G., 28n.

McCann, R. V., 28/2.

McGary, J., 176

McCleary, R. A., 199, 254, 429, 474

McClelland, D. G.3 53, 120, 121, 178,

325 543 710

McClintock, G. G., 423, 463, 473, 474
MacGorquodale, K., 105, 178, 325, 737,

798

808

NAME INDEX

McDougall, W., 13, 20, 53, 260, 293,
301, 369, 383, 427, 432, 474

McGinnies, E., 169

McGrath, J. E.3 421

McHugh, R. B.3 543

McKeel, H. S., 28n.

MacKinnon, D. W., 28rc.

MacLeod, R. B., vii, 6

McNemar, Q., 325

McPherson, D.3 605, 610

McPherson, J., 605, 610

McQuitty, L. L., 270, 288, 289, 290, 300,
325

McRae, D., Jr., 543

Madler, G., 327

Mahalanobis, P. C., 289

Mahler, M., 157

Maier, N. R. F., 146n., 178, 300

Malinowski, B., 45, 53, 620

Marks, G., 597, 610

Marriott, F. H. C., 725, 759, 760, 772,
798

Marshall, A., 620

Martin, D. R., 324

Marx, K., 63, 467, 468, 620

Marx, M., 178, 182, 543

Maslow: A. H.5 21, 53, 194, 196, 254,
300

Masserman, J. H., 145, 178, 300

Mathis, A., 545n.3 606, 610

Maudsley, H., 6 In.

May, M. A., 300, 325

Mayo, E., 421

Mead, G. H., 621, 622, 655, 656, 691

Mead, M., 45

Meehl, P. E,, 54, 105, 178, 262, 270,
285, 325, 737, 798

Meeland, T., 289

Meister, D. E., 176

Mellinger, G. D., 421

Mendel, G. J., 331

Mendelejeff, D. I., 278

Meng, H., 155, 178

Menkes, A., 129n., 178

Menkes, J., 129n., 178

Menninger, K., 248

Merlan, R, 6 In., 178

Merriman, G., 344, 345, 360

Merton, R., 45

Meynert, T., 60, 61

Midlo, C., 358

Mill, J. S., 270, 778

Miller, G. A., 390n., 394, 421

Miller, J. G., 28n.

Miller, N. E, 54, 79, 86, 137, 140, 144,
145, 146, 158, 169, 178, 260, 383,
474, 741, 742, 748, 751, 755, 757,
759, 763, 764, 765-766, 767, 771,
772, 773, 781, 782

Mitchell, F. H., 217, 219, 254

Mittelman, B., 158

Money, J., 360

Moore, M., 28n.

Moreno, J. L., 549

Morgan, G. D., 28n., 54, 137, 178, 260

Morgan, C. T., 751, 771

Morgan, L., 38, 53, 290

Mosak, H., 219, 254

Mosteller, F., 506

Mowrer, O. H., 28n, 54, 79, 86, 137,
140, 144, 145, 146, 147, 158, 178,
256, 300, 325, 383

Mueller, G. G., vi, vii, 6, 759, 775, 789

Muench, G. A., 219, 254

Munroe, R. L., 79, 133, 143, 178

Murawski, B. J., 28n.

Murchison, C. M., 179, 420, 472

Murphy, G., 53, 139, 169, 178, 421, 474

Murphy, L. B., 421

Murray, H. A., 4, 7, 54, 80n., 137, 139,
140, 158, 170, 178, 179, 181, 260,
293, 294, 325, 460, 474, 591, 621,
622, 683n., 691, 745, 750, 751, 762,
763, 768, 769, 771, 780

Nachmansohn, M., 179

Nagel, E., 543

Neuhaus, J. G., 326

Newcomb, T. M., 5, 260, 384, 389'n.,

411, 414, 420, 421, 422, 434, 443,

472, 473, 474, 745, 771
Newman, H. H., 344, 345, 360
Newstetter, W. I., 411, 422
Newton, I., 15, 135, 698, 779
Nunberg, H., 85, 179
Nunnally, J. G., 203, 254

O'Connor, J. P., 219, 254

Odbert, H. S,, 322

Olds, J., 612n., 623, 624, 628, 639, 644n.,

645, 648, 650, 655n., 672, 673, 675,

678, 680, 681, 682-683, 684, 690,

707, 710
Oparin, A. L, 38
Orr, I., 344, 361

Osburn, H. G., 286, 288, 306, 325
Osgood, C. E., 250, 254, 289, 326, 422,

443, 444, 474

Page, J.3 179

Pareto, V., 45, 50, 53, 620, 625

Parsons, A., 612n.

Parsons, T., 5, 45, 53, 157, 543, 612,

655n., 690n., 710, 711, 737, 738n.,

745

Pasamanick, B., 360
Patrick,!. R., 411, 422
Paul, L, 142n.
Pavlov, I. P., 54
Peak, H., 436, 474
Pearl, R., 700, 710
Pearson, K., 259

Perkins, H. V, 547, 597, 605, 610
Perlmutter, H. V., 421
Peterson, D. R., 315

NAME INDEX

809

Petrullo, L., 421

Piaget, J., 69, 96, 109, 122, 123, 124,
1253 1287i., 132, 136, 137, 138, 140,
147, 149, 158, 159, 162, 179, 656,
658, 710

Pirenne, M. EL, 725, 759, 760, 772, 798

Planansky, K., 360

Plato, 38

Poetzl, O., 67, 142, 179

Postman, L., 168, 169, 738, 751

Powder-maker, H., 180

Prell, D. B., 315, 324, 351, 359

Prentice, W. G. H., 768

Prince, M., 13, 53, 70

Pumpian-Mindlin, E., 175, 177

Putnam, M. C., 157

Quinn, R. D., 215, 255

Race, R. R., 360

Radke, M., 242, 255

Rado, S., 133

Raimy, V. G., 194, 201, 217, 219, 254,

255

Rainer, J,, 359
Ramsey, G., 139, 168
Rank, O., 53, 57, 13-3, 187
Rao, G. R., 326
Rapaport, D., 5, 55, 59, 69, 71, 72, 79,

88, 91, 104, 110, 115, 137, 140, 142,

143n., 145, 151, 154, 174, 179, 180,

181, 182, 260, 744, 745, 746, 762,

771, 779

Rappaport, S., 59

Raskin, N. J., 217, 219, 244, 255, 256
Raven, B. H., 459, 474
Razran, G. H., 435, 460, 474
Redfield, R., 50, 54
Redl, F., 104, 181, 548, 610
Rehage, K. J., 605, 610
Reich, W., 157, 181
Reider, N., 144, 181
Reik, T., 62, 133, 181
Reisner, D., 360
Rethlingshafer, D., 281, 326
Rhymer, R. M., 323
Rich, J. L., 8 In., 181
Richard, J., 250
Rickman, J., 173
Riecken, H. W., 28n.
Riesen, A. H., 199
Riesman, D., 157, 441, 474
Ritter, W. E., 17
Robbins, L., 696
Rodnick, E. H., vii, 6
Roe, A., 358

Roethlisberger, F. J., 242, 255, 422
Roffenstein, G., 181
Rogers, B. O., 360
Rogers, G. R., 5, 143, 144, 184, 217, 220,

252, 253, 255, 301, 342, 361, 745,

746, 762, 768, 771

Rogers, N., 219, 256

Roheim, G., 157, 174, 181

Romano, J., 138

Rorschach, H., 137, 139, 181, 281

Rosanoff, A. J., 344, 345, 361

Rosanoff, I. A., 361

Rosenberg, M. J., 439, 474, 542, 543

Rosenblith, W. A., 802

Rosenthal, B., 545ra.

Rosenzweig, S., 28rc., 80n., 137, 181

Rossi, P. H., 543

Rudikoff, E. C., 219, 256

Runkel, P. J, 417, 422

Runyon, D. L., 459, 472

Russell, B., 779

Russell, E. S., 17

Ryans, D. G., 281, 326

Sachs, H., 13

Sander, G., 360

Sanford, R. N., 28n., 139, 181, 472

Sanger, R., 360

Santayana, G., 53

Sapir, E., 45, 53

Sarchet, B., 606, 610, 611

Sarnoff, L, 423n., 463, 473

Satter, G. A., 269, 285, 306, 324

Saunders, D. R., 285, 316, 323, 324, 320

Schachter, S., 410, 414, 420, 422

Schafer, R., 59, 120n., 152, 181, 182

Schanck, R. L., 409, 422, 445, 474

Scheidlinger, S., 548, 610

Scheier, I. H., 315

Schilder, P., 137, 180, 182

Schiller, G. H., 69, 182

Schlesinger, H. J., 109, 176, 182

Schlosberg, H., 176

Schmidl, F., 60n., 6 In.

Schneiderman, E., 325

Schreber, D. P., 157

Schroedinger, E., 127, 182

Schroetter, K., 182

Schull, W. J., 334

Schumpeter, J. A., 625, 703, 711

Schwebel, M., 242, 256

Scott, J. P., 361

Scott, T. H, 168, 175

Scriven, M., 788

Sears, R. R., 140, 142, 147, 158, 182,

326, 373-374, 376, 383, 622, 623
Seashore, S. E., 475
Seeman, J., 194, 215, 217, 244, 256
Sells, S. B., 315
Selye, H., 271
Senden, M., I24n.
Shakow, D., 67, 110, 137, 176, 182
Shand, A. F., 432, 475
Shartle, G. L., 548
Shaw, G. B., 45
Sheerer, E. T., 217, 219, 256
Sheffield, F. D., 462, 473
Sheldon, R. G., 622
Sherif, M., 301, 385

810

NAME INDEX

Shields, J., 351, 361

Shils, E. A., 45, 543, 622, 623, 624n.,

710, 737, 738n.
Shlien, J., 194
Shuttleworth, F. K., 325
Siemens, H. W., 332, 361
Silberer, H., 142, 182
Simmel, A., 479
Simpson, G. G., 80n., 182
Sims, V. M., 411,422
Skinner, B. F., 54, 124n., 341, 342, 361,

365, 375, 383, 711, 725, 738, 750,

759, 771, 778

Slater, E., 333, 341, 345, 351, 361, 624
Slavson, S. R., 548
Smelser, N. J., 649n., 690n., 707, 710
Smith, G. E., 28n.
Smith, M. B., vii, 6, 429, 434, 475
Snyder, W. IL, 253, 256
Snygg, D., 194, 197
Sokal, R. R., 312
Solomon, R. L.3 176
Sombart, W., 620
Somers, R. H., 543
Sorsby, A., 361
Soskin, W., 545n.
Spearman, G. E., 259, 260, 276, 519n.,

539', 541

Spemann, H., 17
Spence, D. P., 176
Spence, K. W., 260, 326, 742, 758, 773,

782

Spencer, H., 620
Spitz, R., 140, 158
Spranger, E., 475
Standal, S., 194, 200, 207-209, 2233 224,

256

Stein, M. I., 28n.
Steiner, I. D., 422
Steinzor, B., 597, 610
Stekel, W., 133
Stellar, E., 176
Stephenson, W., 202, 247, 250, 256, 289,

545n., 610

Stevens, S. S., 54, 474, 802
Stice, G. L., 315, 324
Stock, D., 217, 219, 256, 545n., 571,

605, 606, 610, 611
Stogdill, R. M., 548, 610
Stotland, E., 5, 260, 423, 438, 472, 745,

762, 771
Stouffer, S. A., 457-458, 475, 543, 622,

700, 711

Strandskov, H. H., 361
Strodbeck, F., 678n.
Strom, K., 217, 256
Stromgren, E., 336, 361
Stroud, H., 312
Stumpfl, F., 361
Suchman, E. A., 430, 475
Suci, G. J., 326, 444, 474
Sullivan, H. S., 30, 53, 57, 67, 79, 102,

116n., 133, 143n., 152, 154, 182
Sutton, H., 334, 361

Svehla, G., 422

Swanson, G. E., 420, 421, 472, 473

Symonds, P. M., 326, 427, 475
Szondi, L., 281

Tagiuri, R., 421, 422

Tannenbaum, P. H., 443, 474

Thelen, H. A., 5, 544, 571, 605, 606,
610,611, 745, 751, 771, 780

Thetford, W. N., 219, 256

Thibaut, J., 420

Thomas, W. L, 621, 622, 623

Thompson, H., 331, 359

Thompson, W., 312

Thorndike, E. L., 54, 344, 345, 361

Thorndike, R. L., vii, 6, 290, 755

Thornton, G. R., 281, 326

Thums, K., 352, 361

Thurstone, L. L., 259, 260, 276, 277,
280, 281, 282, 286, 314, 315, 326,
346, 361, 430, 475, 534, 536-537,
539, 541n., 543, 770

Tinbergen, H., 182

Tinbergen, N., 69, 711

Tolman, E. G., 20, 28n., 54, 80, 105,
182, 482-483, 543, 617, 620n., 622,
623, 695; 711, 718, 734, 735, 736,
737, 738, 739, 747, 748, 751, 755,
767, 769, 771, 779

Tomkins, S., 28n.

Torr, D. V., 292, 326

Toynbee, A., 47

Travell, J., 155, 182

Trendelenburg, F., 800

Troland, L. T., 137n., 182

Trotter, W., 45

Tucker, L. R., 286, 326

Urbanek, J., 796

Vandenberg, S. G., 346, 361

Varcndonck, J., 182

Vargas, M., 217, 247-248, 256

Vernon, P. E., 326

Videl, H., 331

Virchow, R., 114

Volkart, E. H., 414, 421

von Bckesy, G., 795, 799, 802

von Bertalanffy, L., 16, 50, 93n.s 168

von Bracken, H., 358

von Frisch, K., 617n.

von Ucxkull, J., 26

von Verschuer, O., 332, 344, 362

Waeldcr, R., 91, 182
Wagman, M., 464, 475
Wald, G., 38, 800, 801, 802
Wapner, S., 122, 177, 183
Ward, J. H., 286, 326
Warkcniin, J., 179

NAME INDEX

811

Warner, L., 537

Warner, W. L., 543

Waterhouse, I. K., 327

Watson, J. B., 10, 54, 293, 427, 475, 623

Watson, W. S., 475

$fe MH"46°79 468, 620, 623, 625,

68871., 691, 711
Weinberg, W., 336, 343, 362
Weiss, E., 155, 183
Weiss, W., 462, 473
Wendt, G. G., 333, 362
Wenger, M. A., 326
Wenig, P., 302, 324, 326
Wernlr, H. 69, 109, 122, 137, 158, 177,

183, 360

Wertheimer, M., 85n., 475
Wexler, M., 157, 183
Wheeler, R. H., 85n.
Wheelis, A., 84n.
White, L., 611
White, M. S., 422
White, R. K., 548
White, R. W., 28n., 429, 434, 475
Whitehead, A. N., 7, 21-29 38 42, 50,

53,624,626,649, 702,711, 779
Whyte, L. L., 38
Wiener, A. S., 362
Wiener, N., 91, 93, 183
Wiggins, L. M., 538, 543
Wilkins, G., 461, 474
Williams, J. R., 301, 327

Wilson, J. T., vii, 362

Wilson, R. N., 28n.

Wineman, D., 181

Wingfield, A. H., 344, 345, 362

Withall J., 546, 547, 597, 605, 611

Wittels, F., 62, 183

Wittenborn, J. R., 290, 327

Wolf, K., 140

Wolfe, D. M., 438, 472

Wolff, P. H.. 138, 183

Wolff, W., 253, 256

Wolfle, D., vii, 1, 6, 770

Woodrow, H., 315, 327, 770

Woodworth, R. S., 88, 96, 183, 439, 475

Wright, S., 335, 362

Wrigley, C., 326

Wundt, W., 364n.

Wyatt, F., 28n.

Yarrow, P. R-, 46 1, 474

Zander, A., 547, 609
Zazzo, R-, 339, 348, 362
Zelditch, M., 624
Zener, K. E., vi, vii, 6
Zetzel, E., 143, 183
Zimmerman, J., 219, 256
Zimmerman, W. S., 280, 325
Zsoldos, J., 84n.

SUBJECT INDEX

Topics followed by an asterisk are those treated by all or most authors (often
extensively). These pertain mainly to the "crosscutting" systematic and methodic
issues raised by the themes of analysis and related editorial proposals. In most in-
stances, page references for such topics are given only to basic definitions or explana-
tions. Individual author treatments of many asterisked topics can be located by
reference to ^the tables of contents appearing with each article, in conjunction with
the use of discussion topic index numbers (see Note on the Use of Discussion Topic
Index Numbers, pp. 724-725).

Ability factors, 273ff., 305
Abreaction, 91

(See also Catharsis)
Academic freedom study, 529
Acceptance, 208, 218-220, 225-226, 431,

453

Accounting equations, 478, 494, 502ff.
Achievement, 452, 652, 664
Achievement concepts, 86, 154
Act-psychology, 61
Action-oriented attitudes, 449, 451-452,

458

(See also Attitude)
Action space, 631, 643-644, 672-673
Actions, 9-12, 74, 375-376, 573-574,

613ff., 646
and attitudes, 431-432, 438, 446, 449,

451-452, 454

(See also Action-oriented attitudes;
Behavioral component of attitude,"
Preliminary theory of attitude
structure and change)
experimental, in thought, 75, 92
normative control of, 630

(See also Control)
primary model of, 71, 72
secondary model of, 73-75
systems of, 623, 628, 630ff.
(See also Theory of action)
Actor, and locus of values, 623

as unit of system of action, 628-629
Adaptation, 69, 73, 74, 78-79, 87-88, 96,
99-101, 115, 153, 449, 631, 633,
638, 644, 664ff., 689
and adjustment, 102
(See also Reality)
Adaptedness, 91, 100-101, 152
Adaptive point of view, 67, 68, 73, 97-

101, 104-105, 154
Additivity in combining factors, 284-285,

299, 305-306

Additivity law in photometry, 795-796
Adequacy, 685-686

Adjustment, 206, 218, 219, 224, 230-

235, 240, 636, 669
and adaptation, 102
in factor analysis approach, 300-301
as stability, 248-249
(See also Congruence; Normality)
Affect, 70-73, 77-78, 97-99, 113, 117,

126, 129, 149, 637-638
charge of, 77
discharge of, 77
discharge channels of, 75, 77
primary model of, 72
as psychological energy, 91
secondary model of, 76-78
signal, 76, 77, 92, 99
Affection, learned need for, 208
Affective associations, 435-436, 449-450,

460, 470-471
Affective component of attitude, 428-

430, 433fL, 444, 446, 470
Affiliation need, 452
"Age of Theory," 731-732

attitude toward generalization range

of laws, 749-752

attitudes re mathematization, 769-770
attrition against, as most pervasive

trend of Study I, 783-785
commitments re observation base, 752-

755

demand for "unambiguous linkage" of
intervening variables to observ-
ables, 743-749

gap between ideology and recent prac-
tice, 786
ideology, re hypothctico-deductive

method, 776-777
re intervening variables, 733-735
position re generality of intervening

variable functions, 739-740
premises of, 733

reason to expect readjustment of
rationale and action, 786-787

813

814

SUBJECT INDEX

"Age of Theory," trends of Study I in-
viting rectification of doctrine of,
785-787
Aggression, 77n., 87, 118-119, 127n.,

153, 376, 442, 469, 636, 669
AG1L, 643-644

(See also Adaptation; Goal attain-
ment; Integration; Pattern main-
tenance)

Aging, 18, 345-350
Alien factors, 591
Alpha situation, 27-28
Ambivalence, 634n.
American Psychological Association, 6

Policy and Planning Board, vi
Analysis, blind, 585-587

experiential (see Experiential analysis)
input-output (see Input-output analy-

sis)

level of (see Level of analysis*)
physical, 794
Rank Pattern, 247
rational, 721
in science, 4

themes of (see Themes of analysis)
Anthropology, cultural, 45-47, 635-636

psychoanalytic study of, 136, 157
Anticipation, 75-77

(See also Expectation)
Anti-intellectualism, 166
Anti-Negro test, latent structure of, 534-

535

Anxiety, 72, 113, 156, 204, 213, 218,
220, 227-230, 283, 297, 481-482,
605

as second-order factor, 277, 282, 305
Apparatuses (see Ego)
Apperception, 27-29, 41
Applied psychology, 365
Apprehension, latent structure of, 529
Appropriateness, concept of, 447-453,

465

Approval, 225-226
Art, as expression of personality, 17

psychoanalytic study of, 136, 150, 156
Assemblies, cell, 131, 650
Association, selective, 415, 420
Attitude, action-oriented, 449, 451- 452,

458

and actions (see Actions)
affective component of, 428-430,

433IT., 444, 446, 470
and attraction, 389, 390, 394, 397™

400, 410-416

balanced, 449, 452, 458-459, 460
and behavior, 453-456

(See also Behavioral component of
attitude; Preliminary theory of
attitude structure and change)
and belief, 384-420, 429-430, 433,

452, 459-460
as cathexis, 30, 389fT.
change, 385-386, 389n., 408ff., 416-
417, 423-471

Attitude, cognitive component, 428—431,

437, 446, 448,450, 459,471
concept of, history of use of, 427-428,

466

conditioning of, 460
definition of, 428-429
determinants of, 389n., 436-443, 464
ego-defensive, 449, 452-453, 463-465,

468-469

ego-instrumental, 436, 440-443, 470
in factor theory, 291-292, 294
frequency of interaction and, 408ff.
informational support of, 434
intellectualized, 449-451, 458-460
isolated, compartmentalized, 433, 456
latent structure of, 495-498, 508-527
and motivation, 426, 434-436, 456ff.
object-instrumental, 436, 438-440,

469-470
and orientation, 389, 390, 429, 431-

432, 648
of others, estimates of, 385-386, 413-

414, 419
perceived discrepancy in, 386, 393-

394, 397ff., 445ff.
political, 500

within family, 385
positional, and afFectivity dimensions

of, 430

proximal, 436-438, 469
toward religion within family, 385
scales, 430, 529, 532, 534-538
to self (see Self)
sign nature of, 430
sociological approach to, 298
structure of, 428-432, 456
study of, 5, 465
typology of, 449-453
and value, 428IT., 432-443, 445, 461-

462

(See also Orientation; Value)
Attitudes, orienting* (see Themes of

analysis )
Attraction, and attitude, 389, 390, 394,

397-400, 410-416
liking as, 399-400,417
negative, 394, 398-399
trust and respect as, 398, 399, 401
Audibility curve, 792
Audition, psychophysics of, 792
resonance theory of, 798-799
theories of auditory mechanism, 798-

799

Auditory discrimination, 793, 794
Auditory system, 794
Authoritarianism, 453, 464
Autia, 282

Autism, 302, 402-403, 413-414
Automatization, 68, 79, 88, 100, 108,

145

Autonomy, 68, 74, 75, 84n., 87-88, 91,
96, 98, 100, 102-103, 106ff., 115,
121ff., 153, 154, 159

SUBJECT INDEX

815

Autonomy, from environment, 98, 101,

102, 154

functional, 74, 88, 96, 97, 122
secondary, 96-97, 100
(See also Ego)
Avoidance learning, 436
Awareness (see Consciousness)
Axiomatization,* 135-136, 141, 506,

5955 601-602, 721-722
(See also Hypothetico-deductive
method* )

Background factor in attitude change,

462

Background factors and orienting atti-
tudes* (see Themes of analysis)
Balance (see Equilibrium)
Balanced attitudes, 449, 452, 458-460
Basilar membrane, 795, 798, 799
Behavior, 10, 21, 75, 82ff., 89-91, 103,
121ff., 125-126, 129, 152, 550-556,
613ff., 645ff.
and attitude, 453-456

(See also Action-oriented attitudes;
Behavioral component of atti- _
tude; Preliminary theory of atti-
tude structure and change)
categorization of, 562-564
characteristics of, 83-88, 91-93
communicative, 386-387, 394fL, 402-

403, 617-618, 629-630
determination of, 62, 78n., 83, 84n.,

87-91, 93-104, 106n.3 114
drive (see Drive)
and environment, 121-124
and external frame of reference, 212,

374-378

as fluidly defined, 754
human minimum, 367-368
in laboratory and real world, 424-425.,

467

and motivation, 121-124
and personality, 85-86, 121-124
purposive, 90, 120, 562-564, 572n.~

573n., 574-575
rational and irrational, 44-9
science of, v, 627, 703, 708
sign, 430, 443, 617
as variable, 109-110
(See also Congruence; Group; Incon-
gruence; Neurosis; Psychosis;
Theory, of action)

Behavioral component of attitude, 429,
431-432, 437-438, 444, 446-447,
470-471
Behaviorism, 10-11, 212, 377, 427, 623,

625, 697
classical, 752
its system of orienting attitudes, 754-

755

obj activist epistemology of, 752, 754,
761, 766

Behaviorism, objectivist epistemology of,
independent and dependent varia-
bles, conditions for legitimacy of,
753
stress against, as major convergence

of study, 755-769
Belief and attitude, 384-420, 429-430,

433, 452, 459-460
Beta situation, 28
Bethel workshops, 570-571
Biological processes, 11, 19-20, 39, 41,

456, 616, 705

metabolism, 11, 15, 23,618
Biological sciences and social sciences,

626-627, 69'8-699
Blind analysis, 585-587-
Blood groups and twins, 332-334
Body-mind problem, 651
Bogardus social-distance scale, 431, 455,

529

Boundary processes, 636-637, 643, 645-
647, 65 Iff., 669-670, 673, 679,
707-708
Breakdown, 228-230

(See also Incongruence ; Maladjust-
ment; Neurosis; Pathology; Psy-
chosis )
Bridging problems, 3

Calculus, dynamic, 296-303, 308
Capacity for therapy, 221
Case histories, value of, 11, 16
Catharsis, 441

(See also Abreaction)
Cathectic meaning, 629, 632, 646
Cathexis, 29-30, 71, 75, 77, 92, 94, 111,
113, 121*., 125-129, 153, 636, 653-
656

attention-, 92
and attitudes, 30, 389fL
binding of, 92, 125-126, 128, 147,

153

counter-, 73, 74, 77, 92, 94
deneutralization of, 127
hyper-, 92
mobile, 71, 75, 77, 94, 126, 128, 129,

147
neutralization of, 92-93, 96-97, 100,

125-128, 147, 153
(See also Energy)
Causal texture, lOln., 108, 121
Causation, contemporaneous, 2 Iff., 308
and genetics, 86-88, 328-330
in multivariate analysis, 265, 270-273
in psychoanalytic theory, 83ff., 105ff.
(See also Definitions; Determinism;
Explanation; Reductionism*;
Theory*)

Cell assembly, 131, 650
Censorship, 66, 70, 94, 102, 111, 113,

146, 156

Centers for group study, 547
Central person, 551
Centrality, 551

816

SUBJECT INDEX

Cerebral dysrhythmia in twins, 353
Cerebral palsy in twins, 352-353
Change, 14-18, 43, 308-310, 315, 554-
555, 576, 577ff., 595, 631-632, 635,
639, 642, 671, 688-690
in attitudes, 385-386, 389n., 408ff.,

416-417, 423-471 ^
Characteristics of human infant, 222-

223

Choice patterns, 549
Circular reaction, 96, 109, 123, 147
Class structure, social, 537

(See also Socioeconomic status)
Classification, 694
of factors, 273-278
inferential, 477-478, 480ff., 489ff.,

504-506, 701-702
and measurement, 490
multidimensional, 489, 532
and observation, renewed interest in,

750

physical vs, psychological, 562-564
qualitative, 489, 533
scores, 527, 540

Client-centered therapy, 5, 184-252
compared with psychoanalytic, 143-

144, 248-249
(See also Therapy, in client-centered

framework )

Clinical method, 261-262
Cochlear canals, 799
Codification, 694, 703, 709
Cognition, 35, 138

primary model of, 61, 71, 72
secondary model of, 71, 75-76
Cognitive capacity, 648
Cognitive component of attitude, 428-

431, 437, 446, 448, 450, 459, 471
Cognitive compromise, 457-458
Cognitive map, 444, 617
Cognitive meaning, 379-383, 428-429,

443, 629, 646
Cognitive orientations, 391, 417, 629,

646

Colinearity, 417
Collective properties, latent structure of,

537

Collectivities, 628-629, 649, 654
Color stimuli, 795
Commention, 282
Common sense, 9-11, 137n.
Communication, 157, 160
behavior, 386-387, 394ff., 402-403,

617-618, 629-630, 693
disturbance, 643
persuasive, 407, 412, 415
rewards of, 386, 438
(See also Information theory)
Compensation, 228, 457
Complementarity, 162
Complementary series, 87-88
Complex function, approach to types,
288-290

Complexity as used in classifying factors,

275

Compliance, 103, 155, 441
Compulsions, 228, 671
Concept of the self (see Self)
Concepts,* and conceptions, 137n.
disposition, 480, 484-487
inferential, 480, 487
Lewinian, in Murray's approach, 21-

29

molar vs. molecular, 19
operational, 246, 258, 482-484, 577ff..

598-599

organismic, 17—19
psychoanalytic, 790
self- (see Self)

in sensory psychology, variations in
nature and source of, 791, 798-
801

in social sciences, 477-478, 619
(See also Constructs*; Definitions;
Level of analysis*; Models*;
Theory*)

Concordance in twins, 334ff.
Concreteness, misplaced, fallacy of, 695
Condition response design, 269
Conditional factors, 276, 291
Conditioning of attitudes, 460
Conditions, of reducibility, 512-515
of therapeutic process, 213-215
of worth, 209-210, 224-227, 230, 232
Configural prediction, 287-290, 306
Configuration, of self, in Rogers' theory,

225-226
(See also Self)
as used by Murray, 24
Conflict, 70, 91, 93, 95-96, 111, 117-

118

avoidance and trend toward consist-
ency, 443-444, 457-458
of drive and censorship, 66
of environment and ego, 66
in group (see Group)
as indicated by factor loadings, 286,

296-303

interstructural, 70
in theory of action, 636
unconscious, 112, 153
Conformity, 380-382, 417, 440-441, 460,

463, 469, 635, 686
Congruence, 205-206, 213-221, 228,

230-231, 445, 447
(See also Adjustment; Normality)
Conscious system (see Cs)
Consciousness, 94, 198-199, 221, 229
distortion and denial in, 205, 216,
218, 226, 227, 229, 448, 592
states of, 61, 88-89
(See also Cs)
Consensus, 376-377, 380
Conservation of energy, principle of, 60,

111
Conservation assumptions, 790

SUBJECT INDEX

817

Consistency, 443-44-7, 451, 456-457,

470, 683 ^

(See also Discrepancy; Rationality)
Constraint, 380-382, 447
Construct language,* 718, 719

compared with data language, 117ff.
Construct linkages,* 720
Construction of function forms* (see

Themes of analysis)
Constructs,* hypothetical, 67, 734
in Murray's approach, 20-21
in Rogers' approach, 194-212
(See also Concepts*; Definitions;
Level of analysis* ; Models* ;
Theory*)

Consummatory state, 632, 641, 663
Context and attitudes, 463
Contingency, double, 652-655, 683
Control, conscious and unconscious, 19,

219

co-twin method, 331
of drive, 73, 74, 91, 96, 117, 127-128
hierarchy of levels of, 637, 650, 659-
660, 662, 664, 666, 670-671, 675,
683-685

laboratory, 189, 261-265, 418, 424
Convulsive disorders in twins, 352-353
Correlation, 487, 491
coefficient of, 538
partial, 539

Correlation matrix and higher-order fac-
tors, 276-277

Correlations of factors, 278, 279
Corticalertia, 282
Co-twin-control method, 331
Covariation chart, 268
Covert and overt processes, 10-12, 20,

21, 562-564

Creativity, 16-17, 38-45, 196, 250, 677
Credibility of source and attitudes, 462
Criminal behavior in twins, 353
Criterion rotation, 267
Critical practicality, 282
Cross product, 516, 538
matrix, 519
residual, 524
Cross-sectional studies, 315, 343, 346-

348
Cs (System Conscious), 67, 68, 70, 89,

95, 111

Cultural anthropology, 45-47, 635-636
Cultural differences, 603
Cultural system, 614-615, 617-619, 635,

680
Culture, of group, 551-552, 554, 590

national culture patterns, 289, 603
and personality, 635-636, 657
Curve-fitting, empirical, 721
Curvilincari ty (see Nonlinearity)
Cyclothyme-schizothyme dimension, 280
(See also Manic-depressive psychosis
in twins; Psychosis; Schizo-
phrenia)

Dark adaptation as function of tempera-
ture, 800

Dark-adaptation curves, 797, 800-801
Data, foundation, 732
immediate, 692
objective, limitations of, 374ff., 597-

598

(See also Observer)
psychophysical, 795, 799
relations between physiological and
behavioral, in sensory psychology,
791, 795, 799

sources of, in personality study, 10-11
and systematization, 8, 188, 718-719
Data language,* compared with con-
struct language, 1 1 7fT.
immediate, 718-720
Daydreams, 64
Death instinct, 92n.
Decision making, 585
Decision theory, 798

Defense, 70, 72-74, 77, 87, 91, 92, 94ff.,
99, 117-118, 122, 127, 132, 150,
153, 154, 156, 159, 204-205, 227-
230, 233, 301-303, 306-307, 433,
435ff., 441, 636
and adaptation, 74, 96n.
mechanisms in action theory, 636, 669
and reality, 97, 99
specific mechanisms of, 72, 74, 94, 98,

108rc., 113, 139, 154-155
(See also Aggression; Projection;

Regression; Repression)
Defensiveness, 204-206, 216-218, 226,

228

Deficiency needs, 196, 437
Defining experiments, standard, 482-

483, 736ff.

Definitions, and classification, 489-490
empirical,* 719-720, 744

newer philosophical formulations,

748
relations with other classes of, 747n.~

748?z.

operational,* 119, 246, 258, 471, 482-
484, 577ff., 598-599, 697, 719-
720, 731, 734, 744, 752, 790
partial, 484-487, 744
role of, in systematic formulations, 722
in sensory psychology, 79 1
theory of, for psychological science,

744, 785
Delay, structuralized, 73-78, 92, 94, 99,

114, 125, 127
Demand and supply, 669
Denial to awareness, 205, 216, 226, 227,

229, 592
Density of variable sampling, 278, 305,

318-321
Dependency impulses, 573, 578-579,

581-583

Dependent variable* (see Variables*)
Depth psychology, 13-14
Derivational rigor, 722

818

SUBJECT INDEX

Dermatoglyphic analysis, 332-333
Description and explanation, 378. 56 Iff.,

699

(See also Causation; Definitions; De-
terminism; Meaning; Models*;
Reductionism* ; Theory*)
Desirability, 147

Deteriorating relationships, 236-239
Determinants of attitude, 389n., 436-

443, 464
Determinism, 60, 63, 68, 83, 11 Off., 150,

608
Detour, 73-75, 92, 99, 114, 126, 127

(See also Drive, control of)
Diagnosis, 10, 11, 262, 491, 583-584,

603

Dichotomous system, morphology of, 512
Differential method of estimating twin-
ning, 336

Differential test, 732
Differentiation, 73, 78, 146, 152-153,
640-642, 646ff., 650, 659ff., 671-
681, 685

organic, 616, 646
Diffraction, 794

Dimensions, cyclothyme-schizo thyme, 280
photometric, 795-797
physical, 796-797
Disapproval, 225-226
Discipline, 381, 664
Discomfort-Relief Quotient, 219
Discrepancy, perceived in orientations,

386, 393-394, 397ff., 445ff.
uncertainty regarding, 393-394,

398-400
in verbal expressions and behavior,

453-456, 461
(See also Appropriateness; Incongru-

ence)

Discriminant functions, 289, 333
Discrimination, auditory, 793, 794
psychophysical, 796
visual, 794

Discussion topics, suggested, 713-723
(See also Study of psychology; Proj-
ect A; Themes of analysis)
Disequilibrium, 388, 393-™394, 396-406,

635, 641

progressive, 17-18
Disorganization, 228-230

(See also Discrepancy; Incongrucnce ;
Maladjustment; Neurosis; Path-
ology; Psychosis)
Disposition concept, 484-487
Dispositions to act, 646
Dissociation (see Incongruence)
Distortion in awareness, 205, 216, 218,

226, 227, 448

Distribution of population (see Popula-
tion distribution)
Dizygosity, 332-334
Dominance-submission, 280
Double contingency, 652-655, 683

Dreams, 61, 85, 126, 127, 141, 142

use of, in Murray's approach, 16-17
Drive, 19-21, 39, 60, 61, 66-67, 75, 78,
85, 87, 89-91, 93-98, 106, 107,
122, 132, 153, 686
acquired, 146
activity, 147n.
aggressive, 77«., 87, 118-119, 127w.,

153

(See also Aggression)
and apparatuses, 96
and behavior, 78n.s 85, 89, 91, 95, 98,

106n.
control of, 73, 74, 91, 96, 117, 127-

128

derivative, 71, 73ff., 91, 146
discharge of, 72, 92
discharge threshold of, 71, 73-74,

92ff., 127
and drive object, 71ft7., 77, 90, 91,

98
ego, 70, 89, 94, 102, 111, 113

(See also Ego)
energy, 91-92, 96, 126
gratification of, 71, 73, 75, 96
habit as, 88, 96
hunger, 21, 293, 469, 678n.
libidinal, 37, 69, 111, 127n.
in psychoanalytic theory, 61, 67, 78,

85, 89, 106w.
and reality, 94, 100
representations, 72
theory of, 89-91, 98
(See also Motivation; Motive; Needs)
Drive reduction, 196, 444-445
Drive strength, 298
Drug therapy, 229-230
Dual nature of group, 551-552, 554-557
Dyadic systems, 30-34, 374
Dynamic calculus, 296-303, 308
Dynamic factors, calculus of, 296-303

evidence for, 290-296, 305
Dynamic lattice, 293-297, 301, 307, 313
Dynamic point of view, 67, 68, 70-71,
73, 89-91, 104, 110, 121n., 153

Ear as series of tuned resonators, 798
Economic point of view, 60, 67, 68, 71,
79, 91-93, 104, 111, 121n., 145,
153, 620, 624, 625, 669n., 690, 703,
708
Education, application of Rogers' theory

to, 241-242

Effects of behavior, 19-21
Ego, 66, 70, 77, 78, 94, 98n., 99n., 102,
111, 113, 134, 153, 156, 160, 470,
658, 683
and alter ego in action theory, 652ff.,

669-670
apparatuses of, 94-97. 100, 101n.3

106, 154

conflict-free sphere of, 122, 159
definition of, 95, 99n.

SUBJECT INDEX

819

Ego, development of, 78, 103, 147, 159
and energy, 96, 147
in factor studies, 292, 296
interest, 117, 118
as organ of adaptation, 100
and reality, 66, 99-100
synthetic function of, 95, 99
unconscious, 94, 117, 155
(See also Autonomy; Structure)
Ego defense (see Defense)
Ego-defensive attitudes, 449, 452-453,

463-465, 468-469
Ego-instrumental attitudes, 436, 440-

443, 470

Ego psychology, 61, 66, 68, 79, 89, 91,
97n., 99, 115, 121, 129, 134, 137-
138, 142, 143, 145, 146, 150, 155,
156, 164
Allport's, 145
and therapy, 144n.
Ego strength, 280, 285
Egocentric motivation, 369-373
Embryology, 14-18
Emotion, and conditioning of attitudes,

460

in group behavior, 577ff.
(See also Affect; Feelings; Work-emo-
tionality theory of group as or-
ganism )
Emotionality, 461

(See also Work- emotionality theory of

group as organism)
Empathy, 210-211, 213, 220, 230
Empirical definitions,* 719-720, 744,

747n.-748n.
Empirical vs. systematic variables, 718-

721, 739-740
(See also Variables)
Empiricism, in Murray's approach, 10-

12
in Parson's approach, 626, 692ff., 697,

700ff.

in Rapaport's approach, 82-83
and relation to theory (see Research;

Theory*)
Energy, 67n., 73n., 93, 96, 113, 119,

125ff.

definition of, 16, 308
drive, 91-92, 96, 126
psychological, 91-93, 111, 113, 125ff.
and quantification, 127-129
(See also Cathexis)
Engram, 299, 303, 306
Entropy, principle of, 68, 93, 111, 114,

127
Environment, 72n., 73, 97, lOln., 102,

121ff.5 122n.

average expectable, 100, 154
and heredity, 90-9 1

(See also Psychogcnctic studies of

twins )

human, 619, 645
social, 100-101, 153
(See also Autonomy; Reality)

Epigenesis, 60, 69, 86, 101, 103, 132,

152-154, 159
(See also Ontogenesis)
Epilepsy in twins, 353
Epistemology, 97

Equations, accounting, 478, 494, 502ff.
rational, 721

specification, 266, 286, 287, 305
Equifinality, 80

Equilibrium, 17-18, 34-35, 39, 386-388,
397, 402ff., 625, 631-632, 641, 642,
688-690, 700

Equivalence in factor patterns, 286
Erg, 286, 290-303
Ergic patterns, 294, 306
Ergic tension, 280, 286, 297-298, 307
Esteem, 658, 670
Estimates of attitudes of others, 385-

386, 413-414, 419
Ethics, 381
Ethnocentrism, 408
Ethnology and psychoanalysis, 64, 156
Event, 22-27, 574

macro, proceedings as, 22-27
Evidence for system* (see Themes of

analysis )
Evolution, 14, 18-19, 38-45, 80n., 124,

616, 626
and psychoanalysis, 60, 63, 68-69,

100

social, 45-47, 619
Existentialism, 251
Expectancy rates, life, 336-337, 349-

350

Expectation, 648, 652, 655
(See also Anticipation)
Experience, and attitudes, 435-436
presystematic and systematic analysis

of, 767-768
subjective, 10-12, 188, 191-192,

374ff., 601
as used in Rogers' approach, 197-198,

218-219, 222, 233, 250-251
(See also Observer)
Experiential analysis, presystematic uses

of, 767

revivified concern with, 766-768
some transitional cases, 767-768
systematic uses of, 768
Experimental design, critical, 64n.3 723
in factor analytic approach, 267-270
multivariate vs. univariate, 258-259,

264-265

in Parsons' approach, 694
Experimental method, 131, 261-265,

364, 418, 424ff.

and psychoanalysis, 80n., 130-132,
139-143, 148-149, 156, 161fL,
165-166

(See also Psychoanalysis)
Explanation, 378, 550-5'55, 557-561,
587, 699

820

SUBJECT INDEX

Explanation, types of, for statistical find-
ings, 500-502 f

(See also Causation; Definitions; De-
terminism; Meaning; Models*;
Reductionism* ; Theory* )
Extensionality, 204-205, 216, 218, 219

(See also Intensionality)
External frame of reference, 211-212

(See also Reality)
Extinction, 145-146

Extraversion as second-order factor, 277
Extraversion-introversion, 280, 282

F scale, 463-464, 468

Facilities, generalization of, 662, 666

as inputs, 647-648, 65 Iff., 658, 675-

676

Fact, 692, 699
Factor analysis, 260ff.

compared with latent structure analy-
sis, 538-540

for determining group dimensions,
547-548

loading pattern, 279, 286, 305

model, 265-270, 284ff., 309-310

over time intervals, 265, 268, 3085
310-311

and twin studies, 337

(See also Multivariate quantitative

personality theory)
Factor change, 308-310, 315
Factor order, 318-321
Factor profiles, 289-290, 310
Factors, classification of, 273-278

conceptual status of, 270-273

genetic and environmental, 315

life course of, 315
Faculty psychology, 273
Family life, application of Rogers' theory
to, 241

and homogeneity of attitudes, 385
Family structure, 624, 685
Fantasy, 228, 302, 402

unconscious, 113

use of, in Murray's approach, 16-17
Fear, 382, 435-436, 460
Feedback, 17, 222, 229, 297, 444, 554,
566, 648, 674

(See also Input-output analysis)
Feelings, 9-12, 382, 434ff.

defined by Rogers, 198

expression of, 216, 225-226, 229

(See also Affect; Emotion)
Field concepts, 21-27, 428, 462, 549

(See also Gestalt approach; Lcwinian

psychology)

Fight impulses, 573, 578-579, 581-585
Figural character in attitude change,

462

Fingerprints, 332-333
Fixation, 69, 154

Flight impulses, 573, 578-579, 581-583,
585

Focal length, 795

Folklore, psychoanalytic study of, 136

(See also Mythology; Myths)
Force, and energy, 73n.

psychological, 61, 70, 111, 113, 125,

127

unconscious, 112-113, 118, 153
Force field, 428, 549, 593
Formal organization of system* (see

Axiomatization*; Hypothetic-deduc-
tive method*; Themes of analysis)
Formalization and psychology, 776-782
"Age of Theory" attitudes concerning,

777-778

hypothetico-dedaictive model vs. pre-
scription, distinction, 776
Study I trends concerning, 778-782
(See also Hypothetico-deductive

method*)

Foundation data, 732
Fourier integral, 798
Frame of reference, internal and exter-
nal, 210-212, 461
(See also Reality)
Fraternal twins, 331-334

(See also Psychogenetic studies of

twins )

Free association, 61, 71, 113, 118, 143
Free will, 608
Frequency, 799

concept of, 792-793
definition of, 793, 794
Fourier analysis of, 793, 798
tuning of auditory system, critical

bands in, 795

Freudian theory, 36-38, 55-167, 191,
283, 451, 621, 653-655rc., 658, 680,
683, 692
compared with client-centered theory,

143-144, 248-249
and factor analysis, 292
(See also Psychoanalysis; Psychoana-
lytic theory)

Frustration, 375-376, 437, 655
Fully functioning person, 234-235
Function, change of, 69, 96n.
Function-form specifications,* 720-721
Function forms, in Cattell's approach,

311

construction of* (see Themes of anal-
ysis)

of Lcwin, 121
in psychoanalysis, 121-124
of S-R theory, 121
Functional autonomy, 96, 97, 439
Functioning, pleasure in, 96
Functions, methodological and preposi-
tional, 598-600

"g" factor, 276-278, 539

Gene structure and factor analysis, 312

General factors, 273ff.

SUBJECT INDEX

821

General intelligence factor, 276-278
General system theory, 50-52

(see also Theory, of action)
General theory of action, 612-709
Generalization, as adaptation, 633
of facilities, 662, 666
logic of, 749-750
of meaning, 617-618, 661
Generalization range, problems of, 749-

752

Genetic organizers, 17
Genetic point of view, 67-69, 73, 86-88,

91, 104, 122, 123, 145, 154
as treated by Murray, 16
Genetics, human, 328ff.
Genotypes, 343, 344, 471

(See also Psychogenetic studies of

twins )
Gestalt approach, 19, 21-29, 83-85,

261-262, 549

(See also Field concepts; Lewinian
psychology; Perspective on social
psychology)

Global approach, 19, 261-262
Goal attainment, 631-633, 637, 641,

644, 651-653, 66 Iff., 673
Goals, 66 Iff., 677
Gratification, 632, 641, 652, 653, 662-

663, 674-

Group, cohesiveness, 537, 548-549
explanations of formation of, 550-555
interaction postulates, 5 7 Iff.
leadership, 242, 548, 549, 576
morale, 298-299
nature of, 551-552, 554-557, 564,

574-576
norms and attitudes, 385-386, 406,

414, 458-460, 469
phases, 577-581
reference, 414
social and psychological raison d'etre,

55 Iff.
tension and conflict, 242-244, 591-

592

(See also Interaction; Interpersonal
relations ,* Work-emotionality
theory of group as organism)
Group dynamics, 5, 46-47, 242-244,

408ff.5 548-549, 596
(See also Group)
Group factors, 547-548
Group mind, 369-371
Group synergy, 299, 369-371
Group theory of evolution, 46—47
Guilt, unconscious sense of, 77nv 146,

147, 156
Guttman scale, 529, 532

Habits, 88, 96, 292n., 687

and traits, 480-481
Hcmatological analysis in determining
kind of twins, 332-334

Heredity, and culture, 619
and environment, 90-91

(See also Gene structure; Psycho-
genetic studies of twins)
Mendelian, 341

Hierarchic organization, 60, 69-70, 73,
74, 76-78, 89, 92, 97, 106-108,
123, 128-130, 132, 142, 160, 161
(See also Structure)
Hierarchy of control, 637, 650, 659-
660, 662, 664, 666, 670-671, 675,
683-685

Higher-order factors, ^ 2 73ff.
History, psychoanalytic study of, 136
of psychology, early institutionaliza-

tion as key to, 783-784
scientism in, 783-784
of sensory psychology, 690-691
of system in mediating research* (see

Themes of analysis)
Hodological concepts, 21-29

(See also Field concepts; Lewinian

psychology)
Holistic approach, 19
Holzinger's h\ 337
Homeostasis, 17-18, 34-35, 39, 625
Homogeneous class, 498

mixture of, 498, 505
Homoscedasticity, 284
Homosexuality in twins, 352
Hostility, 442,^ 453, 468
Hullian tradition concerning quantifica-
tion, 741-743, 772-773
Human Dynamics Laboratory, 604-605
Human Nature and Conduct,. 480, 481
Human relations training groups, 570fT.,

593, 603

Humor preferences, factoring of, 283
Hunger drive, 21, 293, 469, 678n.
Huntington's chorea, 290
Hyperatomism, 341
Hyperselectionism, 34-1
Hypnagogic phenomena, 64
Hypnosis, 61, 64, 72, 8<4rc., 112, 139,

159, 459

Hypotheses, specific types of, 559-560
Hypothesis of covert-overt proportion-
ality, 302-303

Hypothetical constructs, 67, 734
Hypothetico-deductive method,* 282,
305, 311, 467, 568, 701-702, 713,
717, 721-722, 732, 734, 776-783
attitudes of systematists, who employ,

780-782
who question fruitfulness of, 778-

780

increased gradualism toward use of, as
a major trend of Study I, 778 782
neo-Lewinian attitude toward, 780
(See also Formalization and psy-
chology)

Hypothetico-deductive model vs. pre-
scription, distinction, 776
Hysteria, 61, 155, 283, 676

822

SUBJECT INDEX

Id, 70, 78, 95, 100, 117, 153, 155

in factor studies, 292 296
Ideal self, 200, 218, 219, 44 Iff.
Identical twins, 331-334

(See also Psychogenetic studies of

twins )
Identification, 69, 96, 302, 339, 441,

551, 653-656, 680
and structure formation, 95, 99
Identity, 160

Ideographic approach and factor analy-
sis, 313

Ideology, 103, 432
Illusions, 792

Imagination, 16-17, 40, 42-44
Importance as orientation variable, 392
Incentive as used in classifying factors.

275, 291n.

Incongruence between self and experi-
ence, 203ff., 213, 216, 226-230,
<£o£

(See also Discrepancy; Maladjust-
ment; Neurosis; Pathology; Psy-
chosis)

Inconsistency (see Incongruence)
Incremental R technique., 265, 269, 313
Independent-intervening-dependent vari-
able schema* (see Variables*)
Independent variable* (see Variables*)
Indeterminacy, 484
Index numbers, explanation of, 5, 724-

725
Indicators, 480-481, 486

relations among, 496, 499, 506
Individual, nature of, 220-221
Individual differences, 259, 260, 283,

375-376, 575
Individual and group psychology, 368-

\) / T

Individual systems of orientation, 384-

420

alternative formulations, 406-408
essential concepts, 388-395

communicative behavior, 394-395
orientation, 388-392
system strain, 393-394
^ systems of orientation, 392-393
history and prospects in mediating re-
search, 416-420
introduction, 384-388
major interrelations among con-
structs, 395-406
relevant evidence, 408-416
Individuality, 116n.

Inductive-intuitive hypotheses, 546-547
Inductors, 17
Induration, 18
Inertia, 631, 632, 647
Infancy, 64

helplessness in, 103
Infant, characteristics of, 222-223
Inferential classifications, 477-478,

480ff.3 489ff., 504-506, 701-702
Information, 390ra., 394ff.5 400, 676

Information theory, 167, 773-774, 798
Initial evidential grounds for system*

(see Themes of analysis)
Inner ear, mechanical properties of, 795,

Input-output analysis, 640-642, 644

647-649, 65 Iff., 656-657, 669, 672.
679-681, 689, 707-708
Inquiry, in group behavior, 574-576

use of, in Murray's approach, 10-11
Insight and change in attitudes, 453,

463

Instigation to communicate, 395, 396
Instinct, 20, 223, 686-687

death, 92n.

Institutional influences on attitudes, 385
Institutipnalization, 615, 627
Integration, 17-18, 24, 300-301
in action theory, 63 Iff., 636, 644,

665ff., 681, 683

(See also Adjustment; Congruence)
Integrative authorization, 676
Intellectualized attitudes, 449-451, 458-

460

Intelligence, general, 276-278
and neurosis, 285
variations in twins, 343-350
Intensionality, 205, 216, 227, 232

(See also Extensionality)
Intention, 61
Interaction, analysis, 548, 577-587 605-

606, 682fF.
frequency of, 406ff.
social, 370, 387, 425-426, 464, 617-

619, 629-630
tendencies, 593

(See also Group; Interpersonal rela-
tions; Theory of action)
Interdisciplinary research, 340-341:

355-356
Interest and attitude, 291 n.

in people, 9, 10

Internal frame of reference, 210, 222
Intcrnalization, 98fT., 378, 441 615, 619

650-651, 654-658, 680, 684, 685
(See also Introjccted value)
Intcrpcnetration, 613, 635, 649ff., 654,

684-685
Interpersonal relations, 30-31, 66, 79,

116n., 134, 151-152, 235-240
in projective techniques, 152
(See also Group; Interaction)
Interpretation (see Psychoanalysis; Psy-
choanalytic theory)
Interrupted tasks, 106n.
Interspecies transposability of findings,

limits of, 751
rationale for, 749

Intervening variable* (see Variables*)
Intervening variable function, definition

of, 735rc.

generality of, 739-743
quantitative, 742-743
strategy for constructing, 735-739

SUBJECT INDEX

823

Intervening variable paradigm for theory
"construction," 733-749

Introjected value, 209, 225-226
(See also Internalization)

Introspection, use of, in Murray's ap-
proach, 10-11

Introversion-extraversion, 280, 282

Involutional melancholia in twins, 354

Ipsative and normative approach. 309,
561

IQ in twins, 345-348

Irreversible states, 18

Item in latent structure analysis, 49 Iff.

Item analysis, 494-498

Item difficulty, 531

Joint frequencies (see Response fre-
quencies )

Judgments, accuracy of, 413-414, 417,
562

Knowledge, 664

L data, 275, 280, 282, 305
Laboratory behavior, 424-425, 467
Laboratory control, 189, 261-265, 418,

424

Lamarckism, 60, 69
Language, of description, 117ff., 375,
379, 467-468, 477-478, 565-568,
599-600, 692, 718-719
of physical dimensions, 792
psychoanalytic study of, 136
Latent continuum, 492-493, 509
multidimensional, 532
as a rank ordering, 532
zero point and unit of measurement

of, 514

Latent distance scale, 532
Latent and manifest space, 489-490,

502ff.

Latent parameters, 494
identification of, 516
interpretation of, 509, 531
Latent pattern maintenance (see Pat-
tern maintenance)
Latent structure defined, 494
Latent structure analysis, 476-542
classification problems, 479-490
disposition concept, 484-487
property space, 487-489

manifest and latent, 489-490
trait, 479-484
introduction, 477-479
logical foundation, 491-506
accounting equations, 502-506
item analysis, 494-498
"mixture" phenomenon, 498-502
nine steps, 506-528 ^
promises and limitations, 528—542
distribution in latent space, 532 538

Latent structure analysis, promises and
limitations, factor analysis, 538-
540

test theory, 540-542
trace lines, 529-532
Latent structure formulation,, 5
Latent trace (see Trace lines)
Latent typology, 533
Lattice, dynamic, 293-297, 301, 307, 313
Laughter responses, factoring of, 283
Law, of effect, 669n.
of inertia, 631, 632
of interpersonal relationships, 240
Lawful behavior, 24, 28-29, 187, 249-
250, 313, 365, 484-487. 550, 561,
566, 595-598, 618
and attitudes and beliefs, 385ff.
Laws, of learning, 316, 365

psychological, generalization of, 749-

752

of thermodynamics, 16
Leadership, 242, 548, 549, 576
Learning, 73, 79, 109, 138

application of Roger's theory of, 241-

^ 242

avoidance, 436
and change (see Change)
and epigenesis, 122, 132, 152
as focus of theory of action, 616-619,

655
and structure building, 131-132, 145-

146, 149, 158, 159
Learning theory, 69, 79, 115, 117, 122,

123, 144-147, 249-250, 312
and latent structure, 537
and psychoanalysis, 69, 79, 137-138,

140, 144-147, 149, 158
S-R, 15, 32-33, 67, 86, 107, 109, 110,

121, 124n., 145, 260,274
(See also Models*; Systems; Theory*)
Least effort, principle of, 60, 74, 92, 93,

111, 114

Legitimation, 657, 677
Level of analysis,* 716

in Gattell's approach. 258ff., 285ff.,

311-312
in Katz's and Stotland's approach,

424-427

in Lazarsfeld's approach, 477-478
in Murray's approach, 8-9, 19, 21,

27-29, 45, 47-49
in Parsons' approach, 615-619, 649-

651, 692ff., 702ff.
in Rapaport's approach, 66-67
in Rogers' approach, 245-246
statistical vs. speculative, 186, 257-

258
in Thclen's approach, 560-562, 598-

600

(See also Causation; Definitions; De-
scription and explanation ; De-
terminism; Language; Method*;
Models*; Quantification*; Re-
ductionisrn*)

824

SUBJECT INDEX

Lewinian psychology, 2 Iff., 79, 86, 93,

106n., 121ff., 130-131
Libido, 37, 69, 111, 127 n.
Life expectancy in twins, 336-337, 349-

350
Life properties, 15-20, 38ff., 618, 705-

.706
Limited systematization, 732

Linear-frequency theory, 773-774
Linear traceline model, 506-528
Linearity, 265, 284-285, 305-306, 311,

539
Literature and psychoanalysis, 62, 64,

136, 156

Loading pattern, factor, 279, 286, 305
Local independence, 502, 506, 526, 538
Locus of evaluation, 210, 217-219
Logic of science, 4, 251, 484-487, 705,

752, 776

Logical constructs (see Constructs*)
Logical positivism (see Positivism)
Longitudinal studies, 315, 343, 346-

348, 354-355
Love, learned need for, 208, 225-226

Maladjustment, 204, 221, 226, 238-239
in twins, 350-354

(See also Breakdown; Incongruence;
Neurosis; Pathology; Psychosis)
Manic-depressive psychosis in twins,

353-354
Manifest and latent space, 489-490,

502ff.

Marker variables, 279, 282, 305
Mastery, 664
Mathematics, language of, 565 568, 599,

698, 704

use of, in theory, 769-771
Mathcmatization (see Quantification*)
Matrix, ascending, 517, 539
bordered, 513
cross-product, 519
moment, 511
stratified, 511
Matrix transformation theorems, 304-

305

(See also Latent structure analysis)
Maturation, 86, 87, 152
Maturity, 207, 219, 220, 226-227

(See also Congruence)
Meaning, 374, 4-38, 484-487, 561, 614,
617, 628, 629, 632, 634, 646, 661,
704-705
cognitive, 379-383, 428-429, 443,

629, 646

empiricist criterion of, 744, 752, 754
vcrifiability theory of, 744
(See also Definitions)
Measure theory, 798
Measurement, * 721

of cathcctic orientations, 391
as classification, 490

Measurement,* of course of therapy in
Rogers' approach, 215, 217, 219-
220, 232-233, 235, 250-251
ipsative vs. normative, 309
multiform, in Murray's approach, 11
as organizing principle, 529
and science, 189, 484-487
in social sciences, 477-478, 567
(See also Equations; Experimental de-
sign; Latent structure analysis;
Method*; Models*; Multivariate
quantitative personality theory;
Prediction*; Probability; Quanti-
fication*; Rating technique;
Scaling; Statistics)
Mechanism, 38-39
Mechanisms of personality functioning,

668-669

(See also Defense; Primary process)
Medical practice, 621
Medicine, influence of, in Murray's ap-
proach, 9-12

Memory, 61n., 7Bn., 94, 95, 99, 139, 651
conceptual organization of, 76, 79
drive organization of, 76, 79
Mendelian heredity, 341
Mensurational and quantificational pro-
cedures* (see Mathematics; Meas-
urement*; Quantification*; Themes
of analysis)
Mental interiors, 281
Mental processes of ordination, prospec-

tion, and orientation, 35-36
Meson, 790

Metabolism, 11, 15, 19-20, 23, 618
Metaphysics, 23

in behaviorism, 10
Metapsychology, 79n., 104-105, 110,

11 In., 134
definitions in, 105
points of view of, 104-105, 110, 152
Motatheory, 569, 571
Method,* 150ff., 159, 161ff., 165-166
blind analysis, 585-587
of changing attitudes, 463

(See also Attitude)
clinical research, 138, 141
co-twin control, 331
cross-sectional and longitudinal, 346—

348
experimental (see Experimental

method)

genetic, 328-330
hypothetico-deductive (see Hypothet-

ico-deductive method*)
inductive-intuitive, 546-547
of investigating group behavior, 576-

587 "
multiform, of assessment, 11, 13, 16-

17

multiperson, 417-419
objective vs. interpretive, 141, 552—

553, 562-564
as related to theory, 11 5-1 16, 166

SUBJECT INDEX

825

Method, scientific, 165-166, 257-258,

261-265

(See also Case histories; Observation;
Quantification*; Sociometry;
Subjectivity; Tests)
Methodological problems, contributions

of sensory psychology to, 790—791
significance of sensory psychology for,

789-801
some difficulties in statement of, 789-

790

Methodology,* 82n., 115, 166
general scientific, 715, 716
psychological, 718
(See also Science)

Methods, concepts, principles, valuable
outside context of system* (see
Themes of analysis)
Mind, group, 369-371
Mind-body problem, 651
Miniature systems, 8, 311
Minnesota Studies in the Philosophy of

Science, 788rc.

Minority groups, 220, 453-455
Misperception tests, 302-303
"Mixture" phenomenon, 498-502
Modalities, 100, 154

in classification of factors, 273-278
Models,* 424, 695, 713, 716, 732

biochemical, in Murray's approach, 23
economic, 68, 71, 72, 669n., 695, 696
factor analytic, in CattelFs approach,

265-270, 284ff., 309-310
hydraulic, 295
latent structure, 506-532
mathematical, 260. 318, 424, 467, 494
psychoanalytic, 67ff.
combined, 7 Iff.
Darwinian, 68 -69
Jacksonian, 70-71
primary, 7 Iff., 78
of action, 71, 72
of affect, 72
of cognition, 71, 72
reflex arc, 67-68, 70, 73, 107, 108
secondary, 73ff., 78
of action, 73—75
of affect, 76-78
of cognition, 75-76
in Rogers' approach, 190
S-R, in Murray's view, 32-33
S-R-S. in Parsons' view, 650, 678,

682ff.

in Thelen's approach, 562-564
Modes, 100-101, 154
Molar vs. molecular approach, 19, 628
Mongolian* in twins, 352
Monozyqosity, 332-334
Mood change, 309-310, 596
Moral realism, 658
Morale, 298-299
Mortality rates in twins, 336-337
Motility, 94-96, 101

Motivation, 68, 71, 73?z., 74, 77-78, 90,
97, 106fL, 111, 117, 123-124, 127,
129, 139, 148n., 149, 150, 152,
635, 636, 647, 686-688, 691
in Asch's approach, 368-373
in economic theory, 620-621
in Katz's and Stotland's approach,

426, 434-436, 456ff.
in Rogers' approach, 196, 222
(See also Affect; Drive; Feelings;

Needs)
Motivational commitment, 634-635,

657, 663
Motivational components, reality tested

and wishful, 303, 306-307
Motivational factors, 276, 286, 290-

303

(See also Affective associations; Erg;
Ergic strength; Motivation;
Needs)

Motive, actualizing tendency, 196
in factor theory, 29 In.
profit, in economic theory, 620-621
(See also Drive; Needs)
Motive force, 639, 648, 672ff.
Multidimensional classification, 489, 532
Multivariate method and twin studies,

332, 337
Multivariate quantitative personality

theory, 257-322

calculus of strengths, self-sentiment,
conflict, and integration, 296-
303

conceptual status and factor interpre-
tation, 270-273
definition of approach, 257-261
factor analytic experiment, 265-270
factor classification, 273-278
motivational and dynamic factors,

290-296

present status, 278-284
relationship to two scientific methods,

261-265

summary, 303-318
type prediction, 284-290
variable density and factor order,

318-321
Multivariate and univariate experiments,

258-259, 261-265, 290ff., 304
Murray's scaffold of a comprehensive

system, 7-54
adoption of organisms concept, 17-

19
apologia and acknowledgements, 52-

54
influence, of chemical embryology, 14 -

17

of evolutionists, 38-45
of medicine, 9-12
of psychoanalysis, 36-38
of social sciences, 45-47
of Whitchcad and Lcwin, 21-36
additional concepts, 29-36

826

SUBJECT INDEX

Murray's scaffold of a comprehensive sys-
tem, interest, in origins, 12-14
in overt behavior, 19-21
in systems, 50-52
introduction, 7—9
need for comprehensiveness, 47—49
tolerance of uncertainty, 49-50
Musterwert, 333
Mutuality, 154

Mutually shared fields, 370-374
Mythology, psychoanalytic study of, 136,

156
Myths, use in personality study, 17, 36

National culture patterns, 289, 603
National Science Foundation, vi, vii, 6
National Training* Laboratory, 570-571
Natural sciences, relation to social sci-
ences, 484
Nature, of group, 551-552, 554-557,

564, 574-576
(See also Group)
of the individual, 220-221
Need-disposition, 646, 660, 666, 685-

686

Need reduction, 196, 444-445
Need strength, 298
Needs, 19-21, 31-33, 646
affiliation, 452
and attitudes, 435ff., 470

(See also Attitude)
deficiency, 196, 437
and estimation of size of coins, 459
group, 369fT,, 552ff., 573
individual, 369fL, 555-556
to know, 462

for love and affection, 208
for positive regard, 208, 223-224

(See also Positive regard)
quasi, 25, 31, 75n.

for understanding, 438, 450-451, 456
(See also Drive; Motivation; Motive)
Negroes, scaling attitudes toward, 534-

537

Neo-Hullian tradition, 743
Neo-pragmatism, 731-732
Neurology and psychoanalysis, 60
Neurosis, conditioning theory compared
with psychoanalytic theory, 146—
147
as example of incongruence, 203, 227™

229

and intelligence, 285
in twins, 351-352

(See also Incongruence ; Maladjust-
ment ; Pathology )
Neutrino, 790
Newtonian mechanics, 714
Nondireclive therapy, 5, 184 -252

(See also Therapy in client-cen-
tered framework)

compared with psychoanalytic therapy,
143-144, 248-249

Nonlinearity, 269-270, 284-285, 311.

312, 539

Nonparametric variables, 269
Nonparticipants, 553
Normality, 83, 98, 114-115, 136, 643,

668-669

(See also Adjustment; Congruence)
Normative patterns, 657—658
Normative scoring, 309
Novelty, 15-17, 43
Nurturance, 686

O-A personality test, 314-315
Object, cathexis, 653-656
and locus of values, 623, 638
as unit of system of action, 628-629,

636, 680, 685
Object choice, 69

anaclitic, 102
Object-instrumental attitudes, 436, 438™

440, 469-470

Object-relations, 636, 651-656
Object-relevance as orientation variable,

392-394, 440

Objective tests, factoring, 283, 303
Objectivism, 752

(See also Behaviorism)
Objectivity, 10-11, 29, 257-258, 366,
375ff., 552, 557, 562-563, 597-598,
600-601, 625
Observables, antecedent and consequent,

734
Observation, and classification, renewed

interest in, 750

method of, and theory, 115—116
participant, 64, 116n., 151, 161-162
relativity of, 63
(See also Observer)

Observation base of psychology, 752 769
summary of principal trends, 768-

769
Observer, objective vs. interpretative,

374ff., 552-553, 556, 563, 597-598
of social environment in social re-
search, 447-448
(See also Observation)
Occupation, 687

Ocdipal period, 654, 683, 685-686
Ontogenesis, 60, 68-69, 77, 78, 95, 96,

i22

complementary series in, 87-88
undifTerentiated phase of, 95
(See also Epigenesis)
Open systems, 93n., 167
Openness to experience, 206, 218
Operational definitions,* 119, 258, 471,
482-484, 577fL, 598-599, 697,
719-720, 731, 734, 744, 752, 790
in Rogers' view, 246
Operations, reductive symptoms, 747
Opinionaire method in attitude studies,

291
Optimal adjustment, 234-235

SUBJECT INDEX

827

Order in classification of factors, 273-

278

Ordination, 35-36
Organic processes, 15—16
Organic security, 648-649
Organism, group as, 544-608

in theory of action, 613, 615, 619,

645, 647-651, 680
Organismic concept, 17-19, 85-86
Organismic valuing process, 210, 218,

222, 224, 226-227, 231
Organization, as independent factor in

systems, 626, 634, 654n., 682ff.
of system, formal* (see Themes of

analysis )

Organizers, genetic, 17
Orientation, 384-420, 454-455, 614-

615, 628, 629
cognitive and cathectic, 391, 431-432,

438, 629, 632, 646
patterns of, 633-634
systems of, 392-393
value, 638-639
(See also Attitude; Value)
Orienting attitudes* (see Themes of

analysis)

Orthogonal factors, 266
Oscillator, amplitude and frequency of,

792

Ossification, 79
Other-directed man, 441, 469
Out-groups, 468-469
Outcomes in personality and behavior,

218-220

Overcompensation, 457
Overdetermination, 67, 83-85, 105ff.,

122

Overlearning, 146

Overt and covert processes, 10-12, 20,
21, 562-564

P technique, 265, 268, 270-271, 278,
283-284, 287, 300-301, 306

study of clinical case, 294, 308
Pain and attitudes, 435-436
Pairing impulses, 573, 578-579, 581ff.
Parallel profiles, 267
Parameters, transituational invariance

of, in Estes' theory, 743
Parametric analysis, 306, 688-690

(See also Latent parameters)
Parametric categories in theory of ac-
tion, 638fT.

Paranoid ideas, 228, 280
Parmia, 280, 283
Parsimony, 141, 285-286

principle of, and psychoanalytic post-
diction, 64n.

Part-whole relationships, 550fL, 646
Partial definitions, 4'84-487, 744
Particularism, 637-638

Pathology, 83, 98, 114-115, 136, 643,

668-669

(See also Breakdown; Incongruence ;
Maladjustment; Neurosis, Psy-
chosis )

Pattern, of changes, 671
cultural, 614fT.
ergic, 294

factor loading, 279, 286, 305
of meaning, 629
unitary, 271
variables, 637fT.
Pattern Analysis, Rank, 247
Pattern approach to personality, 270,

272-273, 282
Pattern index, approach to types, 288—

290
Pattern maintenance, 631, 633-636, 644,

661fT., 664fL, 676, 677, 681
Pattern similarity coefficient, 279, 289
PCS (System Preconscious), 67, 68, 70,

89, 95, 111

Penetrance, 334-335, 337
Perception, 67, 94-96, lOlrc., 122, 139,
149, 198-199, 249, 309-310, 393-
394, 397ff., 549, 617, 648, 792, 798
apperceptive, 27—29, 41
and personality, 27-28, 198-199, 218,

226, 249, 309-310, 465
of self (see Congruence; Incongru-
ence; Personality; Self)
veridical, 448

Perception and central process, in-
creased interest in, 764-766
Perceptual defense (see Defense)
Performance capacity, 648
Personality, and art, 17, 136, 150, 156
and behavior, 85-86, 121-124
and culture, 635-636
as evolutionary failure today, 1 1
and organism, 613ff., 619, 645, 647-

651
and perception, 27-28, 198-199, 218,

226, 249, 309-310, 465
Rogers' theory of, 221-233
theory of, 85, 116n.
total, study of, 8, 11, 19, 51
type and social attitudes, 453
variations in twins, 343-354
Personality factors, 305
Personality sphere, 275, 318, 320
Personology (see Murray's scaffold of a

comprehensive system)
Perspective on social psychology, 363-

383
controversy between individual and

group psychology, 368-374
data of, 374-379
introduction, 363-367
question of perspective in general,

367-368

study of social influences, 379-383
Phase sequences, 131, 132
Phases of system process, 641-642

828

SUBJECT INDEX

Phenocopies, 334

Phenomenal field, 197, 211, 321, 392
Phenomenological approach, 464
Phenotype, 344, 424-425, 466, 471
Philosophy, in relation to theory of ac-
tion, 704
of science, 753, 776, 790

recent tendencies toward liberaliza-
tion, 787-788
Phobias, 228, 435-436
Photochemical theory, interaction of
physiological and behavioral data
in, 801

Photochemistry, 799-801
Photometric dimensions, as example of
involvement of organism in defini-
tion of stimulus variables, 795-797
Photometry, additivity law in, 795—796
Phyletic diversification of subjects,

trend toward, 751
Phylogenesis, 60

Physical dimensions, language of, 792
Physicalism, 761

Placenta and identical twins, 332
Pleasure, 648, 651, 653

and attitudes, 435
Pleasure principle, 60, 68, 71, 74, 75,

90, 92, 111, 112, 114, 126, 153
(See also Entropy, principle of)
Political attitudes within family, 385
Political theory, 624, 7Q6n., 708
Polity, 649n., 706n.
Polydactyly, 335

Population distribution, 509, 531, 532
as discrete classes, 523, 533
multidimensional, 532
specification of moments of, 514, 521-

523

Positional and parametric change, 639fT.
Positive regard, 207-209, 216-219, 230-

231

need for, 208, 223-224
Positivism, views of, 717, 720, 731-732,

752, 776, 787
Gattell's, 311
Lazarsfcld's, 484-487
Murray's, 11
Rogers', 251

Postdiction in psychoanalysis, 63 66
Postulates,* 734

of work-emotionality theory, 571,

601-602

Potency, 638-640, 672
Potentialities, 18
Power, 662
Praxernia, 280
Preconscious system (Pcs), 67, 68, 70,

89, 95, 111
Prediction,* 424-425, 432, 466 467,

557-561, 607, 715, 716
in action theory, 638, 692fT.
of behavior in Rogers' theory (see

Measurement* )
of change, 309-310, 452

Prediction, in clinical practice, 262
configured, 287-290, 306
conservatism re limits of, 751-752
and postdiction in psychoanalytic ap-
proach, 63-66

in probability terms, 49-50, 566, 608
of social behavior, from nonsocial

settings, 365

from verbal expressions, 453-456
type, from source traits, 284-290
use of inquiry in, 10
(See also Equations; Method*; Prob-
ability; Qauntification*; Statis-
tics)
Preferences, changes of, 437

radio, latent probabilities for, 533-534
sociometric, 549
Prejudice, 447, 449-450, 454-455, 461,

468
Preliminary theory of attitude structure

and change, 423-471
general approach, 424-428
outline, 428-464

affective component, 429-430
assumptions about attitude change,

456-464
attitudes and value systems, 432-

443
behavior and expression of attitudes,

453-456

behavioral component, 431-432
cognitive component, 430-431
concept of appropriateness, 447-453
principle of consistency, 443-447
relation to themes of analysis, 465-471
background factors and orienting

attitudes, 466-468
barriers blocking general theoretical

advances, 471

independent, intervening, and de-
pendent variables, 468—471
summary, 464—465
Premsia, 282
Press, 26-29, 31-33, 591
Pressure, social, 380-382
Prestige, 443, 639
Primary process, 60, 78, 79, 89, 91-92,

99, 114, 126, 128, 141, 153
mechanisms of, 75-76, 92, 111, 112,

126, 141, 153-154
quantification of, 128-129
Principle of consistency, 443-447
Principles of Psychology, 479-480
Privacy, 8Qn., 161
Private and public attitudes, 445
Probability, 481, 487, 492ff., 566
conditional, 521
inverse, 525
latent, 493-494

manifest (see Response frequencies)
Probability theory, 798
Probands, 336-337, 343
Proceedings as macro events, 22-27

SUBJECT INDEX

829

Process, ^15-17, 20-25, 604, 677-679
of action (see Theory of action )
analysis in psychological systems, 645-

647

boundary (see Boundary processes)
conditions, of improving relationships,

239-240

and outcome of deteriorating rela-
tionships, 236-239
conscious and unconscious, 112
latent structure of, 537-538
of reintegration, 230-231
therapeutic, conditions of, 213-217
Process problems, 556, 680-681
Productivity, 584-585
Profile similarity coefficient, 310
Profit motive, 620-621
Programmaticity,* of Cattell's theory,

313-314

of Rogers' theory, 249-250
(See also Themes of analysis)
Project A (see Study of psychology,

Project A)
Projection, 27-28, 228, 302-303 441-

442, 453, 463, 468

Projective tests, 17, 71, 79, 86, 120, 129,
137, 139, 219, 233, 281, 301-303,
463, 468

interpersonal relations in, 152
Properties, of life, 15-20, 38fT. 618

705-706

of units and objects, 639
Property space, 487-490, 502ff.
Propositions in theory building, 569,

572-576, 602
Pretension, 282

Proximal attitudes, 436-438, 469
Psychoanalysis, 5, 11-14, 36-38, 306,

548-549

anthropomorphism in, 83
general psychological theory of 58

115-116, 134, 148, 154-157, 163
influence of, 157ff.
and neurology, 60, 66, 112n., 155
and nondirectiveness, 143-144 248-

249

and normality, 83, 98, 114-115, 136
obstacles in development of, 161 167
and Placet's theory, 138, 140, 147,

149, 159, 162
postdiction in, 63-66
and psychology, 79, 80nM 83, 137ff.,

153, 158
schools of, 57, 102-103, 116n., 133-

134, 143-144
sources of incompatible data for, 143—

148
special clinical theory of, 58, 78, 115-

116, 134, 148, 154-155
validation problems in, 64-66, 80,
84n., 106, 114n., 141-143, 147-
149

(See also Freudian theory; Psycho-
analytic theory)

Psychoanalytic theory, 55-167

achievements and convergence with

other theories, 155-158
background factors and orienting

attitudes, 59-82
construction of function forms, 121-

124

evidence for system, 140-149
formal organization of system, 133-

136
history of system's research mediation,

138-140

initial evidential grounds for assump-
tions, 110-121
introduction, 57-59
methods, concepts, principles of broad

application, 149-155
problem of quantification, 124-133
range of applications, 136-138
structure of system, 82-110
tasks for future, 159-167
Psychogenetic studies, 5, 328-357
of twins, 328-357

ascertainment and analysis of twin

samples, 334-338

determination of zygosity, 332-334
general methodological principles,

330-332

intellectual and personality varia-
tions, 343-350
place of psychogenetics in science,

328-330

procedural limitations and advan-
tages, 338-343
prospects for future research 354-

356
psychopathological variations, 350-

354

schematic recapitulation, 356-357
Psychological adjustment (see Adjust-
ment)

Psychological environment, 26, 27
Psychological laws, generalization rans?e

of, 749-752
Psychological maladjustment (see

Maladjustment )

Psychological orientation types, 667-668
Psychological system and social system,

614, 644-647, 649, 679ff.
Psychology,* act-, 61
and axiomatization, 135
developmental, 140
of ethics, 381
gestalt, 110

history of (see History)
Lewinian, 79, 86, 93, 106n.s 121ff.,

130-131

multiperson, 417-419
nature and nurture in, 90-91
observation base of, 752-769
operational definitions in, 119
and psychoanalysis, 79, 80n. 137ff.

153, 158
and quantification, 93n.,

830

SUBJECT INDEX

Psychology, schools of, 84, 110, 166
as science, 81—82
sensory, 6, 789-801
social, 140

(See also Perspective on social

psychology)

Psychopathic personality, 671
Psychopathological variations in twins,

350-354

Psychophysics, complex interplay be-
tween data and theory, 799-801
Psychosexuality, 61, 69, 73, 89, 153, 154

infantile, 89-90

Psychosis, 64, 156, 157, 160, 228-230
cyclothyme-schizothyrne dimension of,

280

in twins, 353-354
Psychosocial point of view, 66, 67, 101—

104, 134, 164

conception of reality in, 100-101
and dissident schools of psychoanalysis,

103

role of tradition in, 103-104
Psychosomatics, 136, 138, 155
Psychotherapy (see Psychoanalysis;

Psychoanalytic theory; Therapy)
Public and private attitudes, 445
Punishment, 380-382, 551
Purpose, 90, 120, 550-555, 562-564,
572n.»573n., 574-575

Q data, 274, 280-282, 296, 305
a sort, 200, 233, 545ra., 557, 584
Q technique, 202, 246, 274, 288, 289
Qualitative functions,* 720-721
Quantification,* in "Age of Theory,"

734
increased realism and gradualism re

prospects, 770-775
in latent structure analysis, 478, 506-

528

multivariate vs. univariate, 258-259
in Parsons' approach, 696-700
in Rapaport's approach, 79-82, 124fT.,

130, 167

in Rogers' theory, 246-247
in sensory psychology, 791, 798
of systematic relationships, 769-775
attitudes toward, of nonmathcmatical

systematists, 771-772
factor analysis, 775
Hull-type program, medications of,

741-742, 772-773
information and linear-frequency

theory, 773-774

latent structure analysis, 774-775
lessons of sensory psychology, 775
Lewin's program, modifications of,

773

stochastic models of learning, 774
trends among mathematically
oriented systematists, 772-775

Quantification,* in Thelen's approach

565-568, 600-601

(See also Equations; Latent structure
analysis; Measurement*; Meth-
od*; Models*; Multivariate
quantitative personality theory;
Prediction*; Probability; Scaling;
Statistics)

Quantificational and mensurational pro-
cedures* (see Mathematics; Meas-
urement*; Quantification*; Themes
of analysis)
Quantitative techniques and specificity,

721

as inversely related to empirical gen-
erality, 771-772

in sensory psychology, range of ap-
plication, 798
Quasi need, 25, 31, 75n.
Questionnaire technique, and factor

analysis, 281
in social research, 483, 557

R technique, 265, 268, 270-271, 274,

278, 283, 288, 294
Radex theory of factor structure, 319
Radio listeners, latent probabilities for

preferences, 533-534
Random time processes, theory of, 798
Range of application of system* (see

Themes of analysis)
Rank ordering, 489, 532
Rank Pattern Analysis, 247
Rating scales of attitudes, 430
Rating technique, of analyzing group
interaction, 577-587, 598, 600

and factor analysis, 281
Rationality, 438, 449-451, 657

(See also Consistency; Discrepancy)
Rationalization, 96n., 228, 302, 402,
456-457, 459

(See also Defense)
Reaction formation, 302

(See also Defense)
Readership, repeated observations of,

537-538
Reality, defense against, 97, 100

and drive, 94, 100

ego and, 66, 99-100

external, 66-67, 74, 92, 94, 9711., 102,
107-108, 115, 121-122, 1"4

to infant, 222-223

psychological, 74, 97

psychosocial conception of, 100 101

relations to, 78-79, 156

and secondary process, 98, 99

seeking, 574, 578

social, 79, 99, 103-104

and structure formation, 99

(See also Adaptation; Environment)
Reality principle, 60, 74, 99
Reality testing, 61, 94, 99, 111
Recruitment pattern, 525-527, 534

SUBJECT INDEX

831

Reducibility conditions, 512-515
Reduction sentence (see Partial defini-
tions )
Reductionism3* 19, 21, 366, 378, 618.

623-626, 649
genetic, 115, 122, 328-330
(See also Causation; Definitions; De-
terminism ; Explanation ;
Theory*)
Reflex-arc model (see Topographic

point of view)

Regression, 60, 69, 127w., 154, 663
Reification, 651
Reinforcement, 653

of conformity and deviance, 635
(See also Goal attainment; Pleasure;

Reward )
Reintegration, 230-231

(See also Congruence)
Rejection, 225-226
Relational reference of action theory,

614, 622-623
Relativity theory, 714
Relevance as orientation variable, 392—

394
Religion, attitude toward, within family,

385

psychoanalytic study of, 136
and scientific attitude, 14, 186
Repression, 202, 453, 463, 468, 666

(See also Defense)
Research, on attitude change, 465
barriers, 163-167, 316-318
convictions, 595-596
difficulties in twin studies, 338-343
economy of, in multivariate analysis,

264

experience as base of, 601-602
on group behavior, 576ff.
interdisciplinary, 340-341, 355-356
process-centered and organism-
centered, 313

psychoanalytic, training for, 163-164
and theory, 187-188, 244, 249-250,
424-427, 483-484, 558-560, 568-
569, 587ff.
(See also Method*)
Researcher roles, 558-560
Resistance, 144, 150, 154
Respect, 398, 399, 401
Response, as admissible dependent var-
iable, 753
and external frame of reference, 211,

212, 679-681

trends toward reanalysis of, 755-759
Response capacity, 648
Response frequency, 492
joint, level of, 510, 540
random error in, 498
restrictions on, 512
Response pattern, 491-492
Response variables (see Variables*)
Retinal image, 795

Reward, 380-382, 439ff., 551, 641, 651,

655, 658

external vs. internal, 687-688
Rigidity, 205, 227, 232, 285, 449, 453
Rituals, use in personality research, 17
Rivalry in twins, 339
Role, 378, 548, 691
Role change, 309-310, 458, 463
Role relationships, 655, 656, 686-687
Rorschach test, 219, 281
Rotation to simple structure, 316

Salient variable similarity index, 279,

281
Sampling methods in twin studies, 334—

338

Sanctions, 382, 445, 655, 658, 687
Satisfaction, 442, 666
Scaling, 430, 5(29, 532, 534-538, 542,

600
problems in psychoanalytic approach,

128ff.

(See also Rating technique)
Scapegoat, 442, 453
Schemata, 109, 123, 147
Schizophrenia, and psychoanalytic

theory, 160
in twins, 353-354
Schizo thyme factor, 280
School performance and twins, 339
Science, biological and social, 3, 626-

627
development of, v, 8, 81, 189, 257-

258, 485-487
general methodology of, 82n., 135,

165, 715, 716, 754
logic of, 4, 251, 484-487, 705, 752
of man, v
methods in personality and social

psychology, 261-265, 364
and parsimony, 64n., 141, 285-286
and religion, 14, 186
of science, 4, 732
social, concepts, 477-478
Study's perspective on, 4
unity of, 47-49, 624-626, 708
Science and the Modern World, 624-625
Scientific method, 165-166, 257-258,

261-265

Scientism, 783-784
Scope of system* (see Themes of

analysis )
Scoring, ipsative vs. normative, 309

latent, 540-541

Second-order factors, 273ff., 292
Secondary process, 60, 78, 79, 89, 91-
92,94, 98-100, 111, 114, 126, 141,
153, 156
dependence on external stimulation of,

154

integrative role of, 85
Security, 685-686
organic, 648-649

832

SUBJECT INDEX

Selectivity in action theory, 618
Self, 160

in Cattail's approach, 301, 307-308
in Rogers' approach, 196, 200ff., 216,

223ff.
Self-actualization tendency, 196-197,

223
Self-concept and value systems, 433-434,

440ff., 447
Self-interest doctrine, 371-373, 620-621,

691

Self -report in measuring attitudes, 291
Senescence in twins, 346—350
Sensory psychology, 6, 789-801
history of, 690-691
quantitative methods in, 798
some characteristics of, 791—801
variations in nature and source of

concepts in, 798-801
auditory concepts as example of,

798-799
photochemical concepts as examples

of, 799-801
variety of definitions in, of response

variables, 797

of stimulus variables, 791-797
Sentence completion, 582-583
Sentiment, 286, 290-303, 307-308, 432
calculus of, 298ff.
as cathcxis, 30
Sequential analysis, 577-587
Serials, 34-35
Sex, behavior and the self-concept, 230

in Freudian theory, 37-38
Sex differences, and adjustment in twins,

352

and intelligence in twins, 348
Shrewdness, 282
Sign behavior, 430, 443, 617
Similarity, attitudinal perceived, 384-

420

Simple structure, 267, 279, 281, 319
Single-species preoccupation, shift away

from, 750-751
Situation, dimensions of, 287, 305, 308-

310, 321
states and system states, 628, 632-633,

639ff.
Sixteen Personality Factor Questionnaire,

277, 280, 296

Size of coins and needs, 459
Size constancy, 792
Slater's discriminant function, 333
Sleeper effect, 462
Slip of the tongue, 116ff.
Social compliance, 103, 441
Social distance scale, 431, 455, 529
Social Influences, 379-383
Social interaction (sec. Interaction)
Social norm (see Group)
Social object as unit of system of action,

_ 619, 628-629, 653-656, 680, 685
Social psychology (see Perspective on
social psychology)

Social sciences, and biological sciences,

626-627

concepts of, 477-478, 619
and natural sciences, 484
Social structure, class, 537

(See also Socioeconomic status)
Social system vs. psychological system,

614, 627
Socialization, 36, 219, 386rc., 624, 653-

656, 685-686
Socioeconomic status, and estimation of

size of coins, 459
latent structure of, 492, 529, 537
Sociology, 45-47, 624, 626-627, 690-

691

Sociometry, 549, 605
Sodium pentothal therapy, 229-230
Somatic compliance, 103, 155
Source trait, 271-273, 284-290, 305
Space, 487-490, 502ff., 549
Specialization of function, 16

(See also Differentiation)
Specific factors, 273ff.
Specification equation, 266, 286, 287,

305
Stability as inherent property, 16, 18,

631-632, 635, 671, 688-690
(See also Equilibrium)
Standard defining experiment, 482-483,

736ff.

Statistics, correlation (see Correlation)
discriminant functions, 289, 333
explanations of findings, 500-502
and individual differences, 259
ipsative and normative approach, 309,

561

Q sort, 200, 233, 545w., 557, 584
Q technique, 200, 202, 246, 274, 288,

289
R technique, 265, 268, 270-271, 274,

278, 283, 288, 294
sequential analysis, 577-587
(See also Equations; Factor analysis;
Latent structure analysis ; Multi-
variate quantitative personality
theory; Quantification*; Popu-
lation distribution; Prediction*;
Probability; Scaling)
Status (see Socioeconomic status)
Stereotypes, 448-449, 468
Stimulation, 673-674
Stimulus, 97, 98, 123

as admissible independent variable,

753

convergence of treatment as between
S-R and personality-social theory,
762-763

deprivation, 72, 108, 159
equivalence, 76n.
and external frame of reference, 2113

212, 679

in perception psychology, 760-761
"physical energy" criterion df, 753

SUBJECT INDEX

833

Stimulus, and related variables in person-
ality and social formulations, 761—764
as requiring perceptual specification,

756-757, 763-765
and response, as dependent on per-
ceptual identification, 768-769
as terms involving no guarantee of

semantic significance, 768-769
trends toward reanalysis, 755-764
as specified in terms of inferred mean-
ing, 756, 757, 769
as treated, by Gibson, 760-761

in sensory psychology, 759—760
trends toward reanalysis of, 755-764
Stimulus-response (S-R) formulation.
67,86, 107, 109, 110, 121, 124n.',
145, 260, 274

in Murray's view, 15, 32-33
Stimulus-response-stimulus paradigm,

650, 654n., 678, 680, 682ff.
Stimulus variables (see Variables*)
Strain, 388, 393-394, 396-406, 574-

576, 689

toward consistency, 634
(See also Consistency; Disequilibrium)
Strategy for development of system* (see

Themes of analysis)
Stress (see Disequilibrium; Strain)
Stress syndrome, 167
Structural point of view, 67, 68, 73, 85,

89, 93-97, 104, 153
Structure, 24, 60, 66, 71, 74-75, 77-79,
93ff., 94?2., 95?z., 99, 107-108, 115,
121, 126, 127, 129, 152-154n., 159,
641-642

building of, 127-129, 146
and identification, 95, 99
and learning, 131-132, 145-146,

149, 158, 159
definition of, 94, 97
and energy, 92-93, 126
and process, 131-132
and quantification, 131-132
and reality, 99, 104
of system* (see Themes of analysis)
(See also Hierarchic organization)
Structures, cognitive, 95, 417

control, 73-75, 77, 87, 92, 126, 153
defense, 74, 75, 92, 95, 96, 108, 126,

153
means-, 73, 75, 88, 90, 95, 96, 126,

153

as variables, 109
(See also Ego, apparatuses of)
Study of psychology, Project A,1 v— vii
conceptual and systematic ( Study I ) ,

v— vi
grounds for selection of systematic

formulations, 2
themes of analysis, 716—723

Study of psychology, conceptual and sys-
tematic ( Study I ) , themes of
analysis, 716-723
time-reference, 731-732
trends, general (see Trends of Study

I)

for sensory psychology, 789—801
empirical substructure and relations
with other sciences (Study II),
vi, 2-3

panel of consultants, vi, vii, 6
postscript volume, vi, 3, 6
special features, 3—4
steering committee, vi— vii, 6
Styles of teaching, 546-547, 605-606
Subception, 199-200, 205, 206
Subgroups, 583-584

Subjectivity, 10-11, 188, 191-192, 211,
257-258, 375, 546, 557, 597-598,
600-601, 697
Sublimination, 92
Submission, 280
Subsidiation, 294-295, 301
Substitute tasks, 80
Success, 664
Suicide, 351
Superego, 70, 78, 95, 97n., 99-100, 102,

146, 153, 156, 658, 683, 692
in factor studies, 292
Superego strength, 280
Supply and demand, 669
Support, 651-653

Surgency-desurgency, 280, 283, 315
Symbolic logic, 722
Symbolization, 76, 92, 126, 153-154,

198-199, 226, 227, 229, 232
Symbols, in communication, 394—395
distinguished from objects, 455—456
Sympathy, 208
Synaptic resistance, 790
Systematic vs. empirical variables, 718-

721, 739-740
(See also Variables*)
Systematic formulations,* v-vi, 713-723
definition, 1, 713-714
(See also Themes of analysis; Theory*)
Systematization, comprehensive and

limited, 732
postivistic, 717, 720
Systems, action, 613ff., 630fL, 644ff.,
649, 656ff., 673ff., 679ff., 685, 688-
690

basis of, 596-598
biological, 616
dyadic, 30-34
general, theory, 50-52
hierarchical, 17-19, 625-626
interpenetrating, 613, 649-651
miniature, 8, 311
open and closed, 16

1 For a detailed statement of aims, design, working methods, history, etc., of
Psychology: A Study of a Science, see Vol. 1, pp. 1-40. Pages 1-18 comprise the
"General Introduction to the Series"; pp. 19-40, the "Introduction to Study I Con-
ceptual and Systematic."

"834

Systems, of orientation, 392-393
(See also Theory*; Units)

SUBJECT INDEX

T data, 282, 305

T technique, 268, 278

Task problems, 556, 572n.-573n.

TAT, 219, 302

Teacher behavior, 546-547, 597, 605-

606

Temperament factors, 273fT.
Temporal systems, 683
Tension, 71, 121, 152, 572-573, 575-
576, 635

change in therapy, 219, 226

discharge of, 75, 152

increase of, 68, 75, 126, 152

maintenance of, 68, 75, 126, 152

reduction of, 68, 71, 92

and structure, 75
Tension-management, 635
Test, objective, factoring, 283, 303

projective (see Protective tests)
Test theory compared with latent struc-
ture analysis, 540-542
Thema, 31-34

dispositions of, 34
Themes of analysis, 2, 716-723, 735

background factors and orienting atti-
tudes,* 716-717

construction of function forms,* 720-
721

discussion outlines, 4

evidence for system,* 722-723

formal organization of system,* 721-

722

history of system in mediating re-
search,* 722

index numbers, use of, 5, 724—725
initial evidential grounds for system,*

719-720

introduction to, 713-714
mensurational and quantificational

procedures,* 721

methods, concepts, or principles be-
lieved valuable outside context
of system,* 723
programmaticity,* 723
rationale, 714-716
scope of system,* 722
strategy for development of system,*

723

structure of system,* 717-719
for study II, 3
(See also Study of psychology, Project

A)

Theoretical constructs (see Constructs*)
"Theory, Age of" (see "Age of

Theory")

Theory,* accuracy of, 190-191, 561
of action, 612-709

background factors and orienting
attitudes, 619-627

Theory,* of action, general theory and

applications, 627-690
internal differentiation, 672-679
internal structures and processes,

659-671

levels of organization, 681-688
object-relations of psychological

systems, 651-656
psychological and cultural sys-
tems, 656-659

psychological system and organ-
ism, 647-651

structural change, 688-690
summary of problems, 679-681
introduction, 613-619
methodology and scientific signifi-
cance, 690-709
evidential status, 707-709
formal organization, 700-702
function forms, 692-696
initial evidential ground for as-
sumptions, 690-692
mensurational and quantifica-
tional problems, 696-700
scope of application, 702-707
relational reference of, 622-623,

692ff., 708-709

aims of, increased modesty in repre-
senting, 752

attitude, 384-420, 423-471, 476-542
auditory, 799

client-centered, compared with Freud-
ian, 143-144, 248-249
comparative analysis of, 732
construction of, and blind analysis,

585-587

and language, 565-568
(See also Language)
construction assumptions, 592-593
development of, v, 8, 163-167, 257-

258, 699-700
factor analysis, 775

compared with latent structure,

538-540

integration of academic and psycho-
analytic, in Murray's approach,
13-14
latent structure analysis, 774-775

compared with test, 540-542
learning (see Learning theory)
linear frequency and information,

773-774
optical, 794
personality, 7-167, 184-252, 257-321,

328-357
photochemical, differential equations

in, 798

political, 624, 706ft., 708
presystematic, 363
problems, 425-427, 608
psychoanalytic, 55-167
psychological, and culture, 5, 45ff.

SUBJECT INDEX

835

Theory,* quantitative behavioral, 732
and research, 115-116, 166, 187-188,
244, 249-250, 471, 483-484,
558-560, 568-569, 624ff., 692ff.3
697, 700ff.

sensory psychology, 775, 789-801
social psychological, 363-420, 423-

471, 476-542, 544-608, 612-709
stochastic, 774
and systematic formulations, 1, 713-

714

(See also Freudian theory; Level of
analysis*; Method*; Models*;
Reductionism* ; Research; Stim-
ulus-response formulations; Sys-
tematic formulations*; Systems;
Trends of Study I )
Therapy, capacity for, 221

and change in self-concept, 202, 217
client-centered, 143-144, 184-252
in client-centered framework, 184—

252

conclusion, 252
general structure of system, 192-

244

applications, 241-244
constructs developed, 192-212
fully functioning person theory,

234-235
interpersonal relationships, 235-

240

personality theory, 221-233
theory of therapy, 212-221
introduction, 185-192
theory in research context, 244-251
incompatible evidence, 247-249
measurement, 246-247
program and strategy, 249-251
conditions of, 213-215
drug, 229-230
and empathy, 212
psychoanalytic, 55-167

(See also Psychoanalysis; Psycho-
analytic theory)

and regard of therapist for client, 208
successful, index of, 215
Therapy process, 216-217
Thermodynamics, laws of, 16
Thought, 9-12, 139, 382-383

experimental action in, 75, 92, 99
forms of, 76, 93, 126
Threat, 204, 206, 216, 226-228, 230-

231, 398-400
Threshold, absolute, 794

(See also Drive)
Thurstone scales, 534-537
Time, 16, 18, 22, 34-36, 577-578, 638,

677-679"

temporal systems, 683
(See also Aging; Change)
Time-reference of Study I, 731-732
Time series, latent structure of, 537-538
Timidity, 280

Topographic point of view, 60, 67-68,

73, 88-89, 95, 104, 110, 121n.
Topological concepts, 21-29
Trace lines, 493-494. 504-505, 509,

520, 529, 539-540
composite, 520

as conditional probabilities, 521
Training for research, 163-164
Trait, 479-484

source, 271-273, 284-290, 305
and type descriptions, 288, 306
Transcultural factors, 290, 312
Transference, 69, 108, 116, 144, 150-

152, 154

Transformation, 15-18
Transformation analysis, 279, 304-305
Trends of Study I (conceptual and sys-
tematic), general, 5-6, 729-788
concluding perspective, 783-788
formalization and psychology, 776—

782
achieved axiomatic explicitness in

limited area, 780-782 ^
belief that formalization is desir-
able in short-range future, 780
belief that hypothetico-deductive
model represents scientific
practice in incomplete way,
778-780

generalization range of psychologi-
cal laws, problems concern-
ing, 749-752

conservatism re limits of pre-
diction, 751-752
increased modesty of aim, 752
revivified emphasis on observa-
tion and classification, 750
shift away from single-species

preoccupation, 750-751
intervening variable paradigm for

theory "construction," 733-749
generality of intervening variable

functions, 739-743
problem of "unambiguous linkage"

to observables, 743-749
strategy for constructing inter-
vening variable functions, 735—
739

introduction, 730-733
mathematization of systematic re-
lationships, 769-775
"nonmathematical" systernatists,

771-772

systernatists working towards
"strong" degrees of mathe-
matization, 772-775
observation base of psychological

science, 752-769
increased interest in perception

and central processes, 764-766
reanalysis of S and R, 755-764
revivified concern with experien-
tial analysis, 766-768

836

SUBJECT INDEX

Trends of Study I (conceptual and sys-
tematic), general, observation
base of psychological science, sum-
mary, 768-769

for sensory psychology, 789-801
characteristics of sensory psy-
chology, 791-801
concepts, variations in nature

and source of, 798-801
quantitative techniques, range

of application of, 798
response variables, variety of

definitions, 797
stimulus variables, variety of

definitions, 791-797
introduction, 789-791

contributions to methodologi-
cal problems, 790-791
difficulties in statement of
methodological problems,
789-790
(See also Study of psychology,

Project A)

Trust, 398, 399, 401, 417
Twin, senescent, 346-350

study of (see Psychogenetic studies

of twins)

variations, 343-354
Twin-family method, 331, 336-337
Twin-index cases, 335-337
Twin rate, 336
Twin samples, 334-338
Types, 288-290, 306, 667-668
Typological prediction, 287-290, 668

Ucs (System Unconscious), 61, 64, 66-
68, 70, 89, 95, 111, 117, 120, 150,
152

descriptive, 88, 150
dynamic, 88

and psychoanalytic theory, 155-156
Unconditional positive regard, 208, 213,

216, 221, 230-231
and positive self-regard, 209
regard complex, 208-209
Unconscious complex component, 292
Unconscious processes, 11, 13-14, 36,

44, 302-303

and conscious control, 19, 219
and the self-concept, 202, 203ff., 226,

229-230
(See also Ucs)

Unconscious system (see Ucs)
Underlearning, 146
Uniformity, pressures toward, 416
Unitary pattern, 271
Units and systems, 628-629, 637, 639,

642, 646, 681ff.

Unity of science, 47-49, 624-626, 708
Univariate experiments (see Multivari-

ate and univariate experiments)
Universal index, 273, 279, 280ff.
Universalism, 637-638, 658

Valence, as cathexis, 29-30, 389ff.
in group formations, 548-549, 556-

557, 581-583, 590, 594-595
Q sort and, 584

Validation, by blind analysis, 585-587
in psychoanalytic approach, 64—66
Value, 562-564, 585, 629

and attitude, 428ff., 432-443, 445,

. 461-462

as cathexis, 29-30, 632, 646
in ideology and sentiment, 432
introjected, 209, 225-226
locus of, 210, 217-219, 623, 638-639,

659ff.

perceived agreement concerning, 419
relational, 623, 663-665, 668
(See also Attitude; Orientation)
Variable density, 278, 305, 318-321
Variables,* dependent, 718, 721
empirical vs. systematic, 718-721,

739-740

independent, 717-719, 721
independent-intervening-dependent,

as static, 189, 425-426
independent-intervening-dependent

schema, 717-718, 720
individual and collective, 410, 420
intervening, 718, 720, 732, 798

conceived as remote from observa-
tion base, 745
criterion of, "unambiguous linkage"

to observables, 743-749
"defining experiment" procedure for

constructing, 734-739
definitions discrepant with opera-
tional law, 745-747
examination of definitional practice,

746-749

function, definition of, 735ra.
generality of, 739-743
strategy for constructing, 735-

739
minimum significant condition for,

741
paradigm for theory "construction."

733-749

pointer reading for, 736
quantitative, 721, 742-743
as tentative "psychologic," 737
theorists who are reluctant to use,

745

treated at same level as independ-
ents and dependents, 744-745
in Katz's and Stotland's approach,

468-471

in latent structure analysis, 479-487
marker, 279, 281, 305
in multivariate research, 258, 260-

265, 266, 304, 310-311, 745
in Newcomb's approach, 388-395
nonparametric, 269
psychoanalytic, 744-745
behavior as, 109-110
dependent, 58, 105-110, 117, 123

SUBJECT INDEX

837

Variables,* psychoanalytic, external real-
ity as, 107-108
independent, 58, 80, 105-110, 117,

123-124
intervening, 58, 97, 105-106, 117,

123-124
intrapsychic, 80
mathematical, 105
motivations as, 108
psychological vs. physical, 10—11
structures as, 109

response, in sensory psychology, vari-
ety of definitions of, 797
in Rogers' theory of therapy, 220
stimulus, in sensory psychology, 791-

797

proximal, 795

variations in complexity of reduc-
tion to physical terms, 792-794
variations in extent to which ex-
perimenter adopts physical
language, 791-792
variations in extent to which or-
ganism is involved in defini-
tion of, 794-797
color stimulus specification as

example, 795

photometric dimensions as ex-
ample, 795

structure and process, 559
systematic, central trend of contribu-
tors' positions re definition, 747-
749

psychological definition of, 747
vs. empirical definition, 718—721,

739-740, 745^
epistemic levels in analysis of,

739-740
"systems" analogy for interrelations,

745

in theory of action, 639fL, 692ff.
in work-emotionality theory, 587-595
Variation, 661

Vector summation, 297ff., 308
Vegetative processes, 618
Verbal reports, 10-11, 391, 766
Veridical perception, 448
Verstehen, 625
Vicarious function, 80
Visibility curve, 797, 800
Visibility data, 800
Visibility function, 797
Vision, additivity law in photometry,

795-796

angular dimensions in, 794
confusion of saturation and brightness

in, 796
dark adaptation in, 799-800

Vision, psychophysics of, 794

wavelength composition in, 794

Visual angle, 795

Visual discrimination, 794

Visual purple, 799-801
absorption curve of, 800

Visual-purple cycle, retinene and vita-
min A in, 800-801

Visual space3 hypothetical, 798

Vitalism, 38-39

Vulnerability, 203-204, 213, 218, 226

Wald and Clark effect, 800-801

Wealth, 664

Weinberg's differential method, 336

Wholistic factor, 276

Wish fulfillment, 60, 68, 71, 75, 111

Wit, factoring of, 283

Withdrawing vs. aggressing, 669

Work, 555, 561-562, 578ff.

Work-emotionality theory of group as

organism, 544-608
applications, 602-604
background factors, 545-557
construction of function forms., 598—

600

evidence, 607
extensibility and programmaticity,

607-608
formal organization of system, 601—

602
history in mediating research, 604—

607 ^

initial evidential grounds for assump-
tions, 595-598
introduction, 545
mensurational and quantificational

procedures, 600-601
orienting attitudes, 557-569

comprehensiveness of empirical ref-
erence, 564—565
degree and mode of quantitative

specificity, 565—568
formal organization, 568—569
level of analysis, 560—562
models, 562-564
prediction, 557—560
structure of system, 569-595

independent, intervening, and de-
pendent variables, 587-595
postulates and propositions, 571—
576

Zeitgeist, 63

Zygosity determination, 332-334

£ m

116262